{ "cells": [ { "cell_type": "markdown", "id": "c000", "metadata": {}, "source": [ "# Learned embeddings against CellProfiler\n", "\n", "[JUMP-Lite](../datasets/jump_lite.ipynb) {cite:p}`Munoz_2026` measured the same 1,536 wells six ways: four\n", "learned embeddings, the same architecture as one of them with untrained weights, and `cp_measure`, a\n", "CellProfiler-equivalent feature set. Swapping the feature block is a controlled experiment rather than a change\n", "of dataset. The images are lossy and there is one field of view per well, and the dataset page says what that\n", "bounds. [Bringing your own embedding](../tutorials/data/embeddings.ipynb) covers loading one and what `var` no longer\n", "tells you." ] }, { "cell_type": "code", "execution_count": 1, "id": "c001", "metadata": { "execution": { "iopub.execute_input": "2026-10-21T00:50:25.669638Z", "iopub.status.busy": "2026-10-21T00:50:25.669224Z", "iopub.status.idle": "2026-10-21T00:50:47.472793Z", "shell.execute_reply": "2026-10-21T00:50:47.471837Z" } }, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ ":9: UserWarning: 30 of 2550 features have no reference values to centre on in at least one group of 'Metadata_Plate' among the rows selected by reference='negcon', because every value there is missing or infinite. Their centre is set to 0 where it is missing and their scale to 1, so the values measured outside the reference rows pass through unnormalized instead of becoming NaN, and are not comparable across groups. They are flagged in var['degenerate_scale'], and pp.feature_select drops them by default; to drop them now, adata = adata[:, ~adata.var['degenerate_scale']].copy().\n", " mt.pp.normalize(classical, method=\"mad_robustize\", by=\"Metadata_Plate\", reference=\"negcon\")\n", ":9: UserWarning: 103 of 2550 features have no spread in at least one group of 'Metadata_Plate' among the rows selected by reference='negcon'. Adding epsilon=1e-18 to their scale multiplies them by up to 1e18, so they dominate every distance downstream. They are flagged in var['degenerate_scale'], and pp.feature_select drops them by default; to drop them now, adata = adata[:, ~adata.var['degenerate_scale']].copy(). A variance or outlier cut does not catch a feature that varies across a plate but is constant among its control wells; the flag does.\n", " mt.pp.normalize(classical, method=\"mad_robustize\", by=\"Metadata_Plate\", reference=\"negcon\")\n" ] } ], "source": [ "import matplotlib.pyplot as plt\n", "import numpy as np\n", "import pandas as pd\n", "import scanpy as sc\n", "\n", "import mantispy as mt\n", "\n", "classical = mt.ds.jump_lite(\"cp_measure\")\n", "mt.pp.normalize(classical, method=\"mad_robustize\", by=\"Metadata_Plate\", reference=\"negcon\")" ] }, { "cell_type": "markdown", "id": "c002", "metadata": {}, "source": [ "## Preparing each block\n", "\n", "The embeddings are normalized per plate and not selected, for the reason on\n", "[Bringing your own embedding](../tutorials/data/embeddings.ipynb). `cp_measure` is a list of measurements, with the usual\n", "pathologies: features that never vary, that duplicate each other, or that run away to enormous values. Holding\n", "the selection identical across blocks would be the unfair comparison, not the fair one. Here is what each\n", "selection step is worth on it." ] }, { "cell_type": "code", "execution_count": 2, "id": "c003", "metadata": { "execution": { "iopub.execute_input": "2026-10-21T00:50:47.474598Z", "iopub.status.busy": "2026-10-21T00:50:47.474422Z", "iopub.status.idle": "2026-10-21T00:50:49.090641Z", "shell.execute_reply": "2026-10-21T00:50:49.089562Z" } }, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ ":13: UserWarning: feature_select flagged none of the 2550 features as selected; uns['mantispy']['feature_select'] says what each operation removed. Every cutoff here is an absolute threshold on the scale pp.normalize left the values on, so check it against that scale; noise_removal's stdev_cutoff is the usual cause.\n", " mt.pp.feature_select(trial, operations=operations, na_cutoff=0.0)\n" ] }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
featureslargest |value|
none1949.0178622.234375
pycytominer default540.0178622.234375
default + drop_outliers526.0456.745605
+ noise_removal0.0NaN
\n", "
" ], "text/plain": [ " features largest |value|\n", "none 1949.0 178622.234375\n", "pycytominer default 540.0 178622.234375\n", "default + drop_outliers 526.0 456.745605\n", "+ noise_removal 0.0 NaN" ] }, "execution_count": 2, "metadata": {}, "output_type": "execute_result" } ], "source": [ "PYCYTOMINER = (\"variance_threshold\", \"correlation_threshold\", \"drop_na_columns\", \"blocklist\")\n", "# feature_select's own default: drop_degenerate, which removes only what normalize flagged, then pycytominer's.\n", "DEFAULT = (\"drop_degenerate\", *PYCYTOMINER)\n", "\n", "selection = {}\n", "for label, operations in {\n", " \"none\": (\"drop_degenerate\", \"drop_na_columns\"),\n", " \"pycytominer default\": DEFAULT,\n", " \"default + drop_outliers\": (*DEFAULT, \"drop_outliers\"),\n", " \"+ noise_removal\": (*DEFAULT, \"drop_outliers\", \"noise_removal\"),\n", "}.items():\n", " trial = classical.copy()\n", " mt.pp.feature_select(trial, operations=operations, na_cutoff=0.0)\n", " trial = mt.pp.subset_features(trial)\n", " selection[label] = {\n", " \"features\": trial.n_vars,\n", " \"largest |value|\": float(np.abs(np.asarray(trial.X)).max()) if trial.n_vars else np.nan,\n", " }\n", "pd.DataFrame(selection).T" ] }, { "cell_type": "markdown", "id": "c004", "metadata": {}, "source": [ "Left alone, `cp_measure` carries values of order 1e5, which survive a robust per-plate normalization because\n", "they are not constant, merely enormous. The pycytominer default cuts the block to about a quarter of its\n", "features and leaves the largest value where it was. `drop_outliers`, which is in `pp.OPERATIONS` and not in the\n", "default set, removes only fourteen more features and brings the largest value down about four-hundred-fold.\n", "`noise_removal` empties the block: its cutoff is an absolute standard deviation, and on the scale\n", "`mad_robustize` leaves no feature falls under it.\n", "\n", "So `cp_measure` gets the standard selection plus `drop_outliers`, and the embeddings get none." ] }, { "cell_type": "markdown", "id": "c005", "metadata": {}, "source": [ "## Comparing six feature sets\n", "\n", "The question is not whether an embedding works, but how it compares with the alternatives on the same wells.\n", "Two readouts:\n", "\n", "- **`tl.map`**: mean average precision. Rank every other profile by similarity to a query and ask whether\n", " its replicates come first. The pairs are defined so that a replicate must come from a *different\n", " laboratory*, which is the cross-site reproducibility question, and copairs gives every compound a\n", " permutation p-value that is then BH-corrected. This is the metric the JUMP benchmarks report\n", " {cite:p}`Kalinin_2025,Arevalo_2024`, and it is the only one of the two that scales: the\n", " similarity-matrix readouts build a dense n-by-n matrix.\n", "- **`metrics.known_relationships`**: of the compounds annotated to act on the same gene, how many end up in\n", " either tail of the similarity distribution?\n", "\n", "Each is measured under three alignments: the principal components alone, `pp.tvn`, and `pp.harmony`, which\n", "ranked top-three in every scenario of the batch-correction benchmark {cite:p}`Arevalo_2024` and is the last\n", "step of the JUMP recipe. `pp.harmony` needs the optional extra, `pip install 'mantispy[harmony]'`." ] }, { "cell_type": "code", "execution_count": 3, "id": "c006", "metadata": { "execution": { "iopub.execute_input": "2026-10-21T00:50:49.094243Z", "iopub.status.busy": "2026-10-21T00:50:49.094086Z", "iopub.status.idle": "2026-10-21T00:51:07.298233Z", "shell.execute_reply": "2026-10-21T00:51:07.297523Z" } }, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "Downloading file 'refchem_annotations.parquet' from 'https://cellpainting-gallery.s3.amazonaws.com/cpg0016-jump/source_all/workspace/publication_data/2026_jump_lite/metadata/v1.0/jump_lite_refchem_annotations.parquet' to ''.\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "\r", " 0%| | 0.00/319k [00:00:18: UserWarning: 30 of 2550 features have no reference values to centre on in at least one group of 'Metadata_Plate' among the rows selected by reference='negcon', because every value there is missing or infinite. Their centre is set to 0 where it is missing and their scale to 1, so the values measured outside the reference rows pass through unnormalized instead of becoming NaN, and are not comparable across groups. They are flagged in var['degenerate_scale'], and pp.feature_select drops them by default; to drop them now, adata = adata[:, ~adata.var['degenerate_scale']].copy().\n", " mt.pp.normalize(wells, method=\"mad_robustize\", by=\"Metadata_Plate\", reference=\"negcon\")\n", ":18: UserWarning: 103 of 2550 features have no spread in at least one group of 'Metadata_Plate' among the rows selected by reference='negcon'. Adding epsilon=1e-18 to their scale multiplies them by up to 1e18, so they dominate every distance downstream. They are flagged in var['degenerate_scale'], and pp.feature_select drops them by default; to drop them now, adata = adata[:, ~adata.var['degenerate_scale']].copy(). A variance or outlier cut does not catch a feature that varies across a plate but is constant among its control wells; the flag does.\n", " mt.pp.normalize(wells, method=\"mad_robustize\", by=\"Metadata_Plate\", reference=\"negcon\")\n" ] }, { "data": { "text/plain": [ "{'openphenom': 384,\n", " 'dinov2': 384,\n", " 'dinov2_random': 384,\n", " 'subcell': 1536,\n", " 'morphem': 1920,\n", " 'cp_measure': 526}" ] }, "execution_count": 3, "metadata": {}, "output_type": "execute_result" } ], "source": [ "targets = mt.ds.jump_lite_targets()\n", "\n", "\n", "def align(wells):\n", " \"\"\"The three alignments every block is compared under: none, TVN and Harmony, on the same 50 components.\"\"\"\n", " sc.pp.pca(wells, n_comps=50)\n", " mt.pp.tvn(wells, batch_key=\"Metadata_Source\", use_rep=\"X_pca\", key_added=\"X_tvn\")\n", " mt.pp.harmony(wells, batch_key=\"Metadata_Source\", use_rep=\"X_pca\", key_added=\"X_harmony\")\n", " return wells\n", "\n", "\n", "def prepared(model):\n", " \"\"\"Load one feature set and align it three ways.\n", "\n", " The CellProfiler block gets the standard selection; the embeddings get none, for the reason above.\n", " \"\"\"\n", " wells = mt.ds.jump_lite(model)\n", " mt.pp.normalize(wells, method=\"mad_robustize\", by=\"Metadata_Plate\", reference=\"negcon\")\n", " operations = (*DEFAULT, \"drop_outliers\") if model == \"cp_measure\" else (\"drop_degenerate\", \"drop_na_columns\")\n", " mt.pp.feature_select(wells, operations=operations, na_cutoff=0.0)\n", " return align(mt.pp.subset_features(wells))\n", "\n", "\n", "blocks = {model: prepared(model) for model in mt.ds.JUMP_LITE_MODELS}\n", "{model: wells.n_vars for model, wells in blocks.items()}" ] }, { "cell_type": "code", "execution_count": 4, "id": "c007", "metadata": { "execution": { "iopub.execute_input": "2026-10-21T00:51:07.299827Z", "iopub.status.busy": "2026-10-21T00:51:07.299660Z", "iopub.status.idle": "2026-10-21T00:51:54.020111Z", "shell.execute_reply": "2026-10-21T00:51:54.018786Z" } }, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
mAPsignificant
alignmentX_harmonyX_pcaX_tvnX_harmonyX_pcaX_tvn
model
cp_measure0.0740.0500.02695.032.04.0
dinov20.0670.0310.02970.05.05.0
dinov2_random0.0100.0090.0100.00.00.0
morphem0.0850.0270.03585.03.06.0
openphenom0.0390.0160.0187.00.03.0
subcell0.0750.0270.02270.02.02.0
\n", "
" ], "text/plain": [ " mAP significant \n", "alignment X_harmony X_pca X_tvn X_harmony X_pca X_tvn\n", "model \n", "cp_measure 0.074 0.050 0.026 95.0 32.0 4.0\n", "dinov2 0.067 0.031 0.029 70.0 5.0 5.0\n", "dinov2_random 0.010 0.009 0.010 0.0 0.0 0.0\n", "morphem 0.085 0.027 0.035 85.0 3.0 6.0\n", "openphenom 0.039 0.016 0.018 7.0 0.0 3.0\n", "subcell 0.075 0.027 0.022 70.0 2.0 2.0" ] }, "execution_count": 4, "metadata": {}, "output_type": "execute_result" } ], "source": [ "CROSS_LAB = {\n", " \"pos_sameby\": [\"Metadata_Perturbation\"],\n", " \"pos_diffby\": [\"Metadata_Source\"], # a replicate pair has to cross laboratories\n", " \"neg_diffby\": [\"Metadata_Perturbation\"],\n", "}\n", "scored_profiles = {}\n", "\n", "\n", "def scored(name, wells, rep):\n", " \"\"\"Cross-laboratory mAP on the treated wells, and target recall on their consensus against its own null.\"\"\"\n", " treated = wells[~wells.obs[\"Metadata_Control\"].to_numpy()].copy() # still carries obsm[rep]; use_rep reduces it\n", " mt.tl.map(\n", " treated, use_rep=rep, **CROSS_LAB, null_size=10_000\n", " ) # the default; the note under the table says why not fewer\n", " per_compound = treated.uns[\"mantispy\"][\"map\"]\n", " scored_profiles[name, rep] = treated # kept so the per-compound panels below need not score again\n", "\n", " consensus = mt.tl.consensus(treated, use_rep=rep, method=\"median\", min_replicates=1)\n", " # Chance is measured, not assumed to be 10%; the section on chance below says why.\n", " recall = mt.metrics.known_relationships(consensus, targets, n_permutations=100, seed=0).iloc[0]\n", " return {\n", " \"mAP\": float(per_compound[\"score\"].mean()),\n", " \"significant\": int(per_compound[\"is_significant\"].sum()),\n", " \"target recall\": float(recall[\"value\"]),\n", " \"recall null\": float(recall[\"null\"]),\n", " \"recall p\": float(recall[\"pvalue\"]),\n", " \"source variance\": float(mt.metrics.pc_regression(wells, key=\"Metadata_Source\", use_rep=rep)[\"value\"].iloc[0]),\n", " }\n", "\n", "\n", "comparison = pd.DataFrame(\n", " [\n", " {\"model\": model, \"alignment\": rep, **scored(model, wells, rep)}\n", " for model, wells in blocks.items()\n", " for rep in (\"X_pca\", \"X_tvn\", \"X_harmony\")\n", " ]\n", ")\n", "comparison.pivot(index=\"model\", columns=\"alignment\", values=[\"mAP\", \"significant\"]).round(3)" ] }, { "cell_type": "markdown", "id": "c008", "metadata": {}, "source": [ "`significant` counts the compounds whose mAP survives a BH correction over the 301 tested, which is the number\n", "a screener cares about: how many compounds would you have called. It needs a large permutation null. copairs\n", "cannot return a p-value below one over the number of draws plus one, and `tl.map` warns when the null is too\n", "small for the correction to call anything, so keep the default of 10,000 draws." ] }, { "cell_type": "markdown", "id": "c009", "metadata": {}, "source": [ "Four things to read out of the comparison table.\n", "\n", "**The untrained model behaves like one.** `dinov2_random` is the same architecture with random weights, and\n", "it is the control for the whole comparison. Its mAP sits at the floor under every alignment, because random\n", "projections of an image do not reproduce across laboratories, and no correction manufactures structure out of\n", "noise. Every other block beats it on mAP under every alignment, which is the least a pipeline has to show.\n", "\n", "**Uncorrected, the classical block reproduces and the embeddings barely do.** Look at the `X_pca` column of\n", "`significant`: `cp_measure` calls 32 of the 301 compounds across laboratories with no batch correction at all,\n", "and the learned embeddings call between none and five. After `pp.harmony` four of the blocks call 70 to 95;\n", "`openphenom` calls seven. The embeddings need the correction to reproduce across sites much more than the\n", "named measurements do.\n", "\n", "**Alignment matters as much as the choice of model.** `pp.harmony` lifts the mAP of every block but the\n", "untrained one, by 1.5 to 3.2 times. Under Harmony, the choice of block then moves it again by more\n", "than a factor of two, and the number of compounds called, among the five real blocks, from seven to 95. `pp.tvn` does little for\n", "retrieval: no block calls more than six compounds under it. It adds a few on MorphEm and OpenPhenom, and on\n", "`cp_measure` it cuts the 32 uncorrected calls to four. Neither the model nor the correction is the\n", "small decision.\n", "\n", "**The CellProfiler-equivalent set leads without a correction.** `cp_measure` tops the uncorrected column\n", "and calls the most compounds under Harmony, though MorphEm's mean mAP under Harmony is higher, SubCell's is\n", "level with it and DINOv2's slightly below, and under TVN `cp_measure` falls behind MorphEm and DINOv2.\n", "\n", "Each of those compounds had to be told apart from *other compounds*, across laboratories. The JUMP-Lite\n", "benchmark {cite:p}`Munoz_2026` also scores an easier question, phenotypic activity: do a compound's replicates\n", "stand apart from the negative controls on the query's own plate? `tl.map` computes it with\n", "`mode=\"activity\"`, which keeps the controls in and retrieves against the ones plated beside each query." ] }, { "cell_type": "code", "execution_count": 5, "id": "c010", "metadata": { "execution": { "iopub.execute_input": "2026-10-21T00:51:54.024174Z", "iopub.status.busy": "2026-10-21T00:51:54.023961Z", "iopub.status.idle": "2026-10-21T00:51:59.355647Z", "shell.execute_reply": "2026-10-21T00:51:59.354690Z" } }, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
activity mAPactive
alignmentX_harmonyX_pcaX_tvnX_harmonyX_pcaX_tvn
model
cp_measure0.3280.3010.30296.088.078.0
dinov20.3190.2750.32793.078.099.0
dinov2_random0.1280.1140.1572.02.02.0
morphem0.3190.2690.34598.081.0106.0
openphenom0.2950.2510.26791.070.083.0
subcell0.3030.2610.29088.077.084.0
\n", "
" ], "text/plain": [ " activity mAP active \n", "alignment X_harmony X_pca X_tvn X_harmony X_pca X_tvn\n", "model \n", "cp_measure 0.328 0.301 0.302 96.0 88.0 78.0\n", "dinov2 0.319 0.275 0.327 93.0 78.0 99.0\n", "dinov2_random 0.128 0.114 0.157 2.0 2.0 2.0\n", "morphem 0.319 0.269 0.345 98.0 81.0 106.0\n", "openphenom 0.295 0.251 0.267 91.0 70.0 83.0\n", "subcell 0.303 0.261 0.290 88.0 77.0 84.0" ] }, "execution_count": 5, "metadata": {}, "output_type": "execute_result" } ], "source": [ "def activity(wells, rep):\n", " \"\"\"Phenotypic activity: do a compound's replicates retrieve each other against the controls on their plate?\"\"\"\n", " table = mt.tl.map(wells, mode=\"activity\", use_rep=rep, copy=True).uns[\"mantispy\"][\"map\"]\n", " return {\n", " \"activity mAP\": float(table[\"score\"].mean()),\n", " \"active\": int(table[\"is_significant\"].sum()),\n", " }\n", "\n", "\n", "activities = pd.DataFrame(\n", " [\n", " {\"model\": model, \"alignment\": rep, **activity(wells, rep)}\n", " for model, wells in blocks.items()\n", " for rep in (\"X_pca\", \"X_tvn\", \"X_harmony\")\n", " ]\n", ")\n", "activities.pivot(index=\"model\", columns=\"alignment\", values=[\"activity mAP\", \"active\"]).round(3)" ] }, { "cell_type": "markdown", "id": "c011", "metadata": {}, "source": [ "On activity the blocks are close. `cp_measure` leads without a correction and on mean mAP under Harmony;\n", "MorphEm leads under TVN and is just ahead on the count under Harmony (98 against 96); DINOv2 is near both. Under TVN every\n", "trained embedding calls more compounds active than `cp_measure` does. OpenPhenom has the lowest mean mAP of\n", "the five real blocks under every alignment, and the untrained model calls two compounds under every alignment.\n", "\n", "For nearly every compound, every replicate sits at another laboratory, so both readouts ask a query to find\n", "replicates run elsewhere; what changes is what they have to outrank. Without a batch correction the learned\n", "embeddings tell a compound from DMSO nearly as well as the named measurements do, 70 to 81 compounds active\n", "against 88 for `cp_measure`, and have much more trouble telling one compound from another, where they call\n", "five at most against its 32. Harmony narrows that: MorphEm, SubCell and DINOv2 then reach a cross-laboratory\n", "mAP of 0.085, 0.075 and 0.067 against `cp_measure`'s 0.074, and call 70 to 85 compounds against its 95. That\n", "is what you would see if these embeddings carry more of the laboratory than `cp_measure` does, and their\n", "uncorrected source variance further down (about 0.03 to 0.04, against 0.02) points the same way. OpenPhenom is\n", "the exception: it calls seven even after Harmony.\n", "\n", "The embeddings are used out of the box, without fine-tuning to Cell Painting data, on lossy-compressed JUMP\n", "images with one fixed preprocessing. Treat the ranking as a property of this setup, not of the models." ] }, { "cell_type": "markdown", "id": "c012", "metadata": {}, "source": [ "### How much of the Harmony column is the seed\n", "\n", "`pp.harmony` clusters the wells before it corrects them, and the clustering starts from a random draw, so the\n", "seed moves the result, and so does the order of the rows, which the draw is taken over. The tables above are\n", "one run each. Here is every block under five seeds." ] }, { "cell_type": "code", "execution_count": 6, "id": "c013", "metadata": { "execution": { "iopub.execute_input": "2026-10-21T00:51:59.358440Z", "iopub.status.busy": "2026-10-21T00:51:59.358180Z", "iopub.status.idle": "2026-10-21T00:52:39.397980Z", "shell.execute_reply": "2026-10-21T00:52:39.396572Z" } }, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
cross-laboratory mAPcross-laboratory callsactivity mAPactive
minmaxminmaxminmaxminmax
model
cp_measure0.0710.079951060.3230.33991100
dinov20.0670.07169700.3150.3199193
dinov2_random0.0100.011000.1280.13012
morphem0.0810.08683920.3160.3199798
openphenom0.0350.039370.2910.2958991
subcell0.0720.07567720.3020.3058890
\n", "
" ], "text/plain": [ " cross-laboratory mAP cross-laboratory calls \\\n", " min max min max \n", "model \n", "cp_measure 0.071 0.079 95 106 \n", "dinov2 0.067 0.071 69 70 \n", "dinov2_random 0.010 0.011 0 0 \n", "morphem 0.081 0.086 83 92 \n", "openphenom 0.035 0.039 3 7 \n", "subcell 0.072 0.075 67 72 \n", "\n", " activity mAP active \n", " min max min max \n", "model \n", "cp_measure 0.323 0.339 91 100 \n", "dinov2 0.315 0.319 91 93 \n", "dinov2_random 0.128 0.130 1 2 \n", "morphem 0.316 0.319 97 98 \n", "openphenom 0.291 0.295 89 91 \n", "subcell 0.302 0.305 88 90 " ] }, "execution_count": 6, "metadata": {}, "output_type": "execute_result" } ], "source": [ "spread = []\n", "for model, wells in blocks.items():\n", " for seed in range(5):\n", " mt.pp.harmony(wells, batch_key=\"Metadata_Source\", use_rep=\"X_pca\", key_added=\"X_harmony_seed\", seed=seed)\n", " treated = wells[~wells.obs[\"Metadata_Control\"].to_numpy()].copy()\n", " mt.tl.map(treated, use_rep=\"X_harmony_seed\", **CROSS_LAB)\n", " spread.append(\n", " {\n", " \"model\": model,\n", " \"cross-laboratory mAP\": float(treated.uns[\"mantispy\"][\"map\"][\"score\"].mean()),\n", " \"cross-laboratory calls\": int(treated.uns[\"mantispy\"][\"map\"][\"is_significant\"].sum()),\n", " **activity(wells, \"X_harmony_seed\"),\n", " }\n", " )\n", "pd.DataFrame(spread).groupby(\"model\").agg([\"min\", \"max\"]).round(3)" ] }, { "cell_type": "markdown", "id": "c014", "metadata": {}, "source": [ "Five seeds move `cp_measure` between 95 and 106 compounds called across laboratories, MorphEm between 83 and\n", "92, DINOv2 and SubCell between 67 and 72, and OpenPhenom between three and seven; the untrained model calls\n", "none under any of them. The three tiers of calls survive every seed: `cp_measure` and MorphEm above DINOv2 and SubCell,\n", "OpenPhenom far below. So does the order inside the first tier: `cp_measure` calls more compounds than MorphEm\n", "under every seed, and MorphEm has the higher cross-laboratory mAP under every seed. What does not survive is a\n", "gap of a few compounds between two blocks in one run: DINOv2 and SubCell\n", "overlap on the calls, and on the activity count so do `cp_measure` and MorphEm, at 91 to 100 compounds active against 97 to\n", "98. Read the Harmony column as a range, and compare two pipelines under Harmony over several seeds rather\n", "than one." ] }, { "cell_type": "markdown", "id": "c015", "metadata": {}, "source": [ "### One preprocessing choice moves the CellProfiler result\n", "\n", "JUMP-Lite's CellProfiler preparation includes a rank-based inverse normal transform {cite:p}`Munoz_2026`,\n", "which this page leaves out; `pp.rank_int` computes it. It makes every feature's distribution the same shape,\n", "which can matter a great deal for a block of measurements on wildly different scales. Here is `cp_measure`\n", "both ways. The correlation filter then runs on the transformed values, so a somewhat different set of features\n", "survives, and the table says how many." ] }, { "cell_type": "code", "execution_count": 7, "id": "c016", "metadata": { "execution": { "iopub.execute_input": "2026-10-21T00:52:39.402356Z", "iopub.status.busy": "2026-10-21T00:52:39.401987Z", "iopub.status.idle": "2026-10-21T00:52:45.621454Z", "shell.execute_reply": "2026-10-21T00:52:45.620309Z" } }, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
featurescross-laboratory callstarget recallrecall nullrecall pactivity mAPactive
recipealignment
as aboveX_pca526320.0890.0510.010.30188
X_harmony526950.0790.0550.020.32896
with rank INTX_pca603190.1640.1120.010.28384
X_harmony603550.1490.1090.010.30186
\n", "
" ], "text/plain": [ " features cross-laboratory calls target recall \\\n", "recipe alignment \n", "as above X_pca 526 32 0.089 \n", " X_harmony 526 95 0.079 \n", "with rank INT X_pca 603 19 0.164 \n", " X_harmony 603 55 0.149 \n", "\n", " recall null recall p activity mAP active \n", "recipe alignment \n", "as above X_pca 0.051 0.01 0.301 88 \n", " X_harmony 0.055 0.02 0.328 96 \n", "with rank INT X_pca 0.112 0.01 0.283 84 \n", " X_harmony 0.109 0.01 0.301 86 " ] }, "execution_count": 7, "metadata": {}, "output_type": "execute_result" } ], "source": [ "def with_rank_int():\n", " \"\"\"cp_measure prepared in the benchmark's order: the rank transform before the correlation filter.\"\"\"\n", " wells = classical.copy()\n", " mt.pp.feature_select(\n", " wells, operations=(\"drop_degenerate\", \"drop_na_columns\", \"variance_threshold\", \"drop_outliers\"), na_cutoff=0.0\n", " )\n", " wells = mt.pp.subset_features(wells)\n", " mt.pp.rank_int(wells, by=\"Metadata_Plate\", stochastic=False)\n", " mt.pp.feature_select(wells, operations=(\"correlation_threshold\", \"blocklist\"))\n", " return align(mt.pp.subset_features(wells))\n", "\n", "\n", "ranked = with_rank_int()\n", "recipes = pd.concat(\n", " [\n", " comparison.merge(activities)\n", " .query(\"model == 'cp_measure' and alignment != 'X_tvn'\")\n", " .assign(recipe=\"as above\", features=blocks[\"cp_measure\"].n_vars),\n", " pd.DataFrame(\n", " [\n", " {\"alignment\": rep, **scored(\"cp_measure, with rank INT\", ranked, rep), **activity(ranked, rep)}\n", " for rep in (\"X_pca\", \"X_harmony\")\n", " ]\n", " ).assign(recipe=\"with rank INT\", features=ranked.n_vars),\n", " ]\n", ")\n", "recipes.set_index([\"recipe\", \"alignment\"])[\n", " [\"features\", \"significant\", \"target recall\", \"recall null\", \"recall p\", \"activity mAP\", \"active\"]\n", "].rename(columns={\"significant\": \"cross-laboratory calls\"}).round(3)" ] }, { "cell_type": "markdown", "id": "c017", "metadata": {}, "source": [ "The transform cuts `cp_measure`'s cross-laboratory calls from 32 to 19 uncorrected and from 95 to 55 under\n", "Harmony. Its raw annotation recall roughly doubles, but so does its null, so the gain over chance is modest:\n", "the margin grows from 0.038 to 0.052 uncorrected and from 0.024 to 0.040 under Harmony. It is the trap the\n", "section on chance below warns about: a raw recall that looks better mostly because its null rose with it. Its\n", "effect on activity\n", "depends on the alignment: small uncorrected, and under Harmony enough to drop `cp_measure` from first to fourth\n", "of the five real blocks on activity mAP, behind MorphEm, DINOv2 and SubCell, and from level with MorphEm at the\n", "top of the count to last. Under Harmony it also leaves `cp_measure` calling fewer\n", "compounds across laboratories than MorphEm, DINOv2 or SubCell. So where the CellProfiler block ranks against\n", "the embeddings here depends on a preprocessing step, not only on the features. The tables above keep this\n", "page's recipe; this one is the reason not to read their `cp_measure` row as the CellProfiler result." ] }, { "cell_type": "markdown", "id": "c018", "metadata": {}, "source": [ "### What is chance, really?\n", "\n", "`metrics.known_relationships` counts annotated pairs falling in either tail of the similarity distribution,\n", "and a fixed tail width invites the assumption that a map with no information scores 10%. Here it does not.\n", "The tails hold 10% of all pairs, so pairs drawn at random would land there 10% of the time. The annotated\n", "pairs are not drawn at random: RefChemDB links some compounds to hundreds of genes and most to a few, so a\n", "handful of compounds make up many of the pairs, and chance moves with where each representation puts those\n", "few. So the chance level has to be measured. `n_permutations` shuffles which compound each annotation row\n", "names, keeping every gene's number of compounds and every compound's number of genes, and reports the mean\n", "recall over the shuffles as `null` and the share of shuffles that recall at least as much as `p_value`." ] }, { "cell_type": "code", "execution_count": 8, "id": "c019", "metadata": { "execution": { "iopub.execute_input": "2026-10-21T00:52:45.624669Z", "iopub.status.busy": "2026-10-21T00:52:45.624507Z", "iopub.status.idle": "2026-10-21T00:52:45.792421Z", "shell.execute_reply": "2026-10-21T00:52:45.791500Z" } }, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAA3oAAAG4CAYAAAAT7JXuAAAAOnRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjExLjIsIGh0dHBzOi8vbWF0cGxvdGxpYi5vcmcvgI3uAAAAAAlwSFlzAAAPYQAAD2EBqD+naQAAeUBJREFUeJzt3XdUFFcbBvBn6QLSRFSavbco9pJowN5ix57YokaNUWNQP0vs3cRo7L232KIx9kLsHWNXqhQpwtKWsnu/P8DRDaKwLCysz+8cju6dO3PfubMM++69MyMTQggQERERERGR3jDQdQBERERERESkXUz0iIiIiIiI9AwTPSIiIiIiIj3DRI+IiIiIiEjPMNEjIiIiIiLSM0z0iIiIiIiI9AwTPSIiIiIiIj3DRI+IiIiIiEjPGOk6gIJMpVIhODgYhQsXhkwm03U4RERERESkx4QQiI2NhaOjIwwMPjxmx0QvB4KDg+Hi4qLrMIiIiIiI6BMSGBgIZ2fnD9ZhopcDhQsXBtI72srKStfhEBERERGRHpPL5XBxcZHykA9hopcDb6ZrWllZMdEjIiIiIqI8kZXLxngzFiIiIiIiIj3DEb1coFQqkZKSouswSE+Zmpry5j9ERERE9EFM9LQsLi4OQUFBEELoOhTSU1ZWVnByctJ1GERERESUjzHR0yKlUomgoCCYm5ujaNGiHHUhrZPL5YiMjESxYsVgZMRfXyIiIiJ6P35S1KKUlBQIIVC0aFEUKlRI1+GQHhJCIDIyEqmpqUz0iIiIiChTvBlLLshvI3l+fn4IDg7O1jqxsbG4dOkSIiIiIITA3bt3ce7cOZw7dw7h4eHZ2tbVq1ehUqmyGbVuBAQEICgoSNdhZCq/vbeIiIiIKH9iovcJ2LRpEw4fPpzl+gqFArVq1cKECRPw6NEj/Pjjj+jZsyemT5+OPn364Pz589lqv0+fPkhOTtYgcs35+/vj5cuX2a7n7e2NK1eu5HJ0RERERES5i3O/KINr166hTp062LVrFwCgf//+ePDgAczMzDBy5Ehdh5clW7duhY2NzUfj/W+93r1751GERERERES5h4leLklNTUVoaGiubLt48eIfvD7r6dOnCAkJgUqlwhdffCGVv3r1CsHBwahRowYMDAzw7NkzWFlZwcHBAQBw5coVVK1aFRcvXkRCQgLOnTsHAIiJicGVK1dQrly5DG0pFAo8fPgQ1tbWKFOmjFT++vVr+Pv7o2rVqu+NMTg4GE+ePIGJiQkqVaoEOzs7IH2aqYmJCYyMjNRizawc6Xc6ffToEZycnFCiRAkkJibC19cXlpaWOHfuHGrUqAGFQpGhvffVi4uLg4GBAZydnd+77Q/FSERERESUXzDRyyWhoaFwcXHJlW0HBgZKich/jR8/Hvv27UPJkiUhk8lw+vRpAMAff/yB33//HXFxcahbty52796N5cuXo0GDBvD09AQAeHp64vjx49i5cyeioqIwffp0IP16venTp2cYHbt9+za6dOkCBwcHhIWFoUOHDvjtt99w4cIFdO7cGWXLloWJiQni4+MzxHnjxg0sWbIEqampeP78ObZs2YIWLVpg06ZNuHTpEkJDQ9Vizaz80qVL6Ny5M1xdXeHn54exY8eif//+OHPmDIyMjHD37l0sWLAAoaGhGdqrUqVKhnpnzpyBmZkZxowZ895tT5w4MdNYiIiIiIjyCyZ6eubOnTuYPXs22rdvD2tra6m8VKlSOHHiBFJSUlC6dGkoFIr3rl+qVCnMmjULx48fx6pVq4D0EcQ3o3tv/gWA//3vfxg8eDAaN24MlUqFwYMH46effsLMmTOxceNGdOzYEf7+/qhQoUKGdjp27Ig6derAz88P//77L3755Re0aNHig7G+r3z69OlYsWIFunXrhvDwcFSuXBkjR47EoEGDMkzd/G97R48ezVDvzJkzUv3Mtp1ZLGZmZjk8ekRERERE2sFEL5cUL14cgYGBubbtzGzduhXr1q1D//79kZKSgh07dgAAateuDQAwNjaGvb094uLiYGhoCKVSKa2bWfKXmZCQEBw5cgQnT54EALi6uiI2Nhbh4eGoXr06AKBkyZKwsbHJsO6cOXOwfPlylC5dGgCQkJAgLXtfrJmVh4eHo0aNGgCAokWLwsHBAVFRUdlqLzMf2vb7YmGiR0RERET5hc4TvVOnTuG3335DWFgYqlevjqlTp350yuO5c+ewatUqPHr0COvWrUOdOnXUljdq1Oi9H+S7d++OyZMnAwBmz56NvXv3qi0vV64c9u3bp5X9MjIyynR6ZW4KCAhA06ZN4ebmBi8vL/j4+GRat2LFitiwYQOKFCmCs2fPvjdB+pCOHTvi4cOH6NevHywtLQEAlStXhru7O8aMGYNhw4bh9OnTiI6OzrDuyZMnMWbMGNSsWRPHjx/H2bNnNdhbwMPDA2PGjMHIkSNx+fJlqFQquLi4wNbWFhcvXkTVqlVRs2bNTNv7b72sbJuIiIiIKL/TaaJ38uRJtG3bFtOmTUPDhg3x66+/onHjxvDx8VGbdvguLy8vXL58GZ07d8bu3bul0Z53rV69Wm2kysfHB/3798e0adOkssDAQBQtWhQLFy6UyvThIefLli3Dy5cvYWlpicGDB6Np06Z4/vy5dLMTpE9hNDY2xoABA/D8+XOsWrUK/fv3R5cuXWBgYAB7e3u16ZaNGjWS/l++fHkULVoUADBlyhSsWbMGGzZskJLEc+fOYdasWZg1axZWrVqFnj17olu3bhluVrJ06VLMnj0bd+7cQa9evRAbGwukT4l8X6yZlc+cOROLFy/GihUr4OzsjOPHj8PAwACenp64desWZs6ciXnz5mXa3n/rubq6wsTEBAAy3XZmsRARERER5RcyIYTQVeMNGzZE2bJlsW3bNiB96mCJEiUwceJETJgw4b3rxMfHw8LCAkFBQXBxccHZs2fRrFmzD7YzevRo7N27F4GBgdLdKocNG4aIiIgcjeDJ5XJYW1sjJiYGVlZWUCgU8PX1RenSpTmNj3IF32NEREREn67/5h8forMRvbi4OFy9ehWjRo2SyszMzODu7o7Tp09nmuhZWFhkq52kpCRs374dQ4cOzfBIgitXrqBRo0awtrZG06ZN8cMPP+jFqB4RERERkVZNf/9su9xvN0Y37eoBnSV6L1++hBBCejbZG46Ojjh16pTW2jlw4ABev36NwYMHq5UXLlwY3377Lb744gu8fPkSP//8Mw4ePIhLly5l+oy6pKQkJCUlSa/lcrnW4iQiIiIiItIWnSV6KSkpAABTU1O1clNTU2mZNqxfvx7NmzdH2bJl1crnzJmjdl1VgwYNUL58eezevRt9+vR577bmzp2Ln3/+WWuxERERERER5QaDLNTJFUWKFAEAREZGqpVHRkZKy3LKz88Pp0+fxpAhQzIs++/NM0qXLo1SpUp98C6VEydORExMjPSTW49PICIiIiIiygmdJXolSpSAo6Mjrl27plZ+5coVuLm5aaWNDRs2wM7ODp07d/5o3ZSUFLx69QqFCxfOtI6pqSmsrKzUfoiIiIiIiPIbnSV6ADBkyBCsW7cOfn5+AIAdO3bg0aNHGDRokFRnxowZ6NatW7a3rVKpsGnTJvTv3z/D9NDk5GT8/PPP0rP2kpOT8cMPPyApKQndu3fP8X7lJ8nJyUhNTdV1GDqTkpKi1anAREREREQFgU6fozd58mS8ePECFStWRLFixRAdHY21a9eidu3aUp2AgAA8evRIen348GFMnTpV+vA+ePBgWFpaYtiwYRg2bJhU78SJEwgMDMx02qaRkRFKliwJGxsbhIWFwcXFBX/99Zfa8+MKojf98mZq6pw5c1C8eHG1vskqpVKJxMREIH0087/TXVUqVYbn472vLDExUWd3M128eDHMzMwwZswYnbRPRERERKQLOh3RMzY2xpYtW/Dy5Uv8/fffCAsLUxvNA4Bp06Zh//790usmTZpg06ZN2L59O27fvo19+/Zh06ZN+Oqrr9TW++yzz+Dj44PKlStnaFcmk2Hy5MkICwvD8ePH4efnh3///RdffvllLu5t3pg/fz5Wr16tlW399ddfKF68OOzs7NS2mZycjK+//hqFCxeGi4sLTp48CQDYu3cvrK2tUalSJYSFhQEAHj9+jOHDh2slHiIiIiIiyhqdjui9YW9vD3t7+/cuc3FxUXttZ2cHOzu7j26zePHiKF68+AfrGBgYZLgbZ0EmhEBycjKSk5MRFxeXYRTtfaNtSqUShoaG791e+/btERcXh/Hjx6uVb9myBQEBAXj16hVu3ryJ3r1748WLF1iyZAkePXqE3bt3Y/v27fjhhx/w008/Ye3atZnGnJycDAMDAxgZGWUrvv+Wf2w7Wd3n1NRUKBQKGBsbZ5jyCwAJCQlQqVSwtLR8b9m7cbxZZm5urlb+bvtZ3T8iIiIiouzQ6YgeaZevry/mz5+PiRMnonjx4jh79iwA4ObNmyhXrhzMzc0xa9YsAEBISAhatWqFwoULo0iRIli3bl2W2zlz5gyGDBkCCwsLfP7553B0dMS///4LAwMDhIeH4/Xr1zA2Nsby5cvRo0ePD95Fdc6cORgyZAhKlSqFwoULY9KkSR+Mz8/PD40bN4a5uTlcXFxw6NChD27nXVnZ582bN6N48eKwsbFBtWrV8ODBAwDAunXrYGtri6JFi6J48eJQKpXvLZswYQL27dsnba9KlSpQKBSYM2cOBgwYgDJlysDS0hLr16/HgAEDYG5ujs8++wyvX7/O8XEhIiIiInqDiV4uSw2NQNKD59Lr5Me+SH2ZNq1RpUhC0t3HUMWl3RQm9VUUku4/e1v3WQBSAkMBACIlFUl3H0Mpj8u0rTJlymDKlClYvHgx4uLi4OHhAQAICgrCnTt34Ovri2XLliE1NRUTJkyAh4cHQkNDce3aNSxcuDDDoy4yExkZqTYCa29vj8jISEydOhWdOnXC+fPn0aRJE1y6dAkWFhawtbVF2bJlERAQ8N7t+fv7w8fHB76+vjh06BBu3ryZaXwTJkxAmzZtkJiYiH379mHIkCHSQ+zft513ZWWfBw0ahLi4OERHR2Pq1KnScxM3bNiAQ4cOIT4+HnFxcTA0NHxv2YdER0fj/v37uH37Nr799lt0794dCoUCtWvXxt69e7McIxERERHRxzDRy2XyLYcR4vmj9Dps6HREL98JAFAGhyPIYzCS7j4GAMTuOY7gzqOluq9GzcHrxZvS6kZGI8hjMBRX7mU7ho4dO8LS0lJ6pEVMTAx8fHwwffp0ODo6ombNmnj58qV099OPcXBwQGhoqPQ6JCQEDg4OaNWqFfz9/XHhwgXMmTMHixYtwpIlS3Dnzh2MGzcOW7dufe/22rdvj8KFC8PBwQHu7u548uRJpvE9fvwYvXv3hoGBAerXr49ixYohODg40+28Kyv7vHfvXjg7O8Pa2hoDBw7E8+dpSfrSpUuxZMkSNGvWDD/99BNSUlLeWyaTydS2p1Qq1fbT3NwclSpVgp2dHdq3bw+ZTIbatWsjPDw8yzESEREREX1MvrhGT59Z9e8Ii/ZfSK+LrZkOA0tzAIChY1E4n1oH47Jp1yEW7tEa5s3qSXUdfpsEmalJWt0iNnA+tQ5GpZ0+2J65uTkCAgIQGxsLc/O0dv6bfAghULduXXTr1k26aykAtevOkH5NX0JCAlJSUqTr/iwtLdG6dWssW7YMTZs2xfXr1xEdHY2qVatK623YsAEtWrSAk5MTDA0NER4ejqioqPde8wYAe/bswVdffYWEhAQcO3YMQ4cOzTS+mjVrYvny5fDy8sLly5cREREBZ2fnTLfj7+8vtZOVff7tt98wZ84ctGvXDvv378fvv/8OAKhevTq2bduGhIQEtGvXDt7e3qhfv36GMkdHR5w8eRIeHh44c+YMXr58KW373ePw32sIhRBZjpGIiIiI6GM4opfLjIrbw7TK2xu+mFQsDSOnYgAAAzNTmNasKCV+Rg52MK1W7m3dcq4wdkm7oYzM2AimNSvC0OrDH/o7dOiAv//+G05OTjh79ixMTEzUHotgbm4OmUyGhQsXwtfXF/Xq1cv0xjWPHz9G8eLFsX79ekydOhXFixdHUlISPD098cUXX6Bp06aYP38+du3aJU1bDA8Px4kTJ6THWvzvf/9Dz549cezYMXz99dfvjbl+/frw9PRE27ZtMWrUKFSrVi3T+BYsWABfX19Ur14dM2bMwI4dO6T9e992TExMYGKSlixnZZ+///57TJs2DQ0bNkR4eLiULDds2BDFixdHjRo1ULlyZTRs2PC9Zd988w38/Pzw2Wef4e7du6hUqRJkMlmG42BhYfH2OGczRiIiIiKij5GJN0MJlG1yuRzW1taIiYmBlZUVFAoFfH19Ubp0aZiZmek6vAJh+vTpGj/nLze2k9/xPUZEREQ6Md1aR+3G6KbdfOq/+ceHcESPdOq/I1263g4RERERkT7gNXqkU+97DIIut0NEREREpA84okdERERERKRnmOgRERERERHpGSZ6REREREREeoaJHhERERERkZ5hokdERERERKRneNfNXFbK66hWt+c3r90Hl//++++IiIjA1KlTAQC7du3ClStX8Msvv2g1DiIiIiIiyr+Y6OmZwYMHo0GDBvDw8ICjoyOmT5+Oixcv6josIiIiIiLKQ5y6qWdMTEywadMmDBkyBH379sWvv/6KokWLZlp/2bJlWLBgATp27Ij69etj586dAICEhAR4eXmhTp06KFeuHFavXg0A2LlzJ8qVK4cqVaqgb9++ePXqVZ7tGxERERERZQ1H9PRQjRo1UK1aNYSGhqJVq1YfrBsVFYX9+/dj1apVSEhIwMCBA1G3bl0sX74cQUFBWL16NaytrWFvbw8AaNeuHerWrYvU1FQcPHgQU6ZMkZJAIiIiIiLKH5jo6SFvb288e/YMhQsXxtGjR9Gu3Yev6/v666/RuHFjAMBXX32FGzdu4Ny5c9i7dy/Kly+vVvfx48f4+eef8eLFCygUig+OFhIRERERkW5w6qaeiYmJwZAhQ7B161Zs2rQJY8eORURExAfX8fb2hlKpRGJiIq5evQoXFxeUKVMGf/75Z4a6EyZMQIcOHbB//378/PPPSEpKysW9ISIiIiIiTTDR0zPDhg3D6NGjUaVKFZQqVQpeXl749ttvP7iOgYEBSpUqhWLFiqFatWpo3Lgx5s+fjw0bNqBo0aJq1+i1adMG48aNQ5s2bfDXX3/l0V4REREREVF2yIQQQtdBFFRyuRzW1taIiYmBlZUVFAoFfH19Ubp0aZiZmeV5PCqVCr6+vihbtqxa+fPnz1G6dGkYGGTM66dPn47ixYtj0KBBiI2NhZ2dndryV69eQS6Xw97eHjY2NgCA169fw8jICBYWFnj58iVcXFxyec/oDV2/x4iIiOgTNd1aR+3G6KbdfOq/+ceH8Bo9PWJgYJAhyQOAsmXLYtmyZVi2bJlaeYsWLVCsWDEAgLGxcYYkDwAcHBzg4OCgVmZrayv9n0keEREREVH+w0TvE9G3b1+0bdtWrczS0hImJiYwMuLbgIiIiIhIn/AT/ifCzs7uvSN2RERERESkf3gzFiIiIiIiIj3DRI+IiIiIiEjPMNEjIiIiIiLSM0z0iIiIiIiI9AwTPT2zdetW7N27V3p99epVzJ49O9P6Bw4cwIkTJ/IoOiIiIiIiygu862Zu0/bDJT/y0Mg2bdqgUaNGqFOnDuzt7TFw4EDs3r070/r//vsvbGxs0LJlS+3GSUREREREOsNET8/Y29tj6dKl+Prrr+Hi4oIRI0agWrVq760bFhaGPXv2wNjYGN7e3pgwYQLWrFmDFStWwNDQEAAwZswYzJgxAydOnICZmRnu3r2LwMBADBs2DJ999lke7x0REREREWUFp27qoXbt2sHe3h4vXrzAd999l2k9CwsLVK5cGTVr1sRXX30FR0dHxMfH488//wQAXL9+HT4+PrCyssL9+/cxbtw4FC5cGC4uLujcuXMe7hEREREREWUHEz095Ovri/v370Mul8PHxyfTepaWlqhevTpq164NT09PFC9eHMOHD8eqVasAACtXrsTw4cOl+kOGDMHo0aMxefJkmJubIybmw9NIiYiIiIhINzh1U8+kpqaib9++WLZsGaysrDBgwABcuXIFJiYm761vaGgIlUolvW7UqBFCQ0Nx+/ZtXLx4EWvWrJGWWVpaSv83NjZGSkpKLu8NERERERFpgiN6embGjBmoX78+WrVqhYYNG6JNmzaYMmVKpvWrVq2KxYsXo2fPnrh79y4AYOjQoejcuTP69OkDIyN+F0BEREREVNDwU3xu+8hdMrVJpVKhRo0a6NChw9vmp0/HoUOHoFKpYGCQMa/v2LEjSpQoAV9fXxQrVgwA0L9/f9ja2qJFixZSvW7duqFQoULS6/nz58PKyirX94mIiPIpbd9VOsvt8rIBIqKskAkhhK6DKKjkcjmsra0RExMDKysrKBQK+Pr6onTp0jAzM9N1eGr27duHffv2qZXVrVsX48aN01lMlH35+T1GRJ8YJnpEnxb+zucL/80/PoQjep+ISpUq4auvvlIrc3Z21lk8RERERESUe3R+jd6aNWtQtWpV2Nvbo3nz5rh58+YH68fHx2PdunWoU6cOLC0tcfHixQx1xowZA0tLS7WfevXq5bjtgqxatWrw9PRU+2nSpImuwyIiIiIiolyg00Rv+/btGD16NKZMmYKbN2+iatWqcHd3R3BwcKbrzJw5E5cvX8b//vc/xMfHQ6lUZqijUCjQokULhIaGSj/nz5/PcdtEREREREQFgU4Tvblz5+Kbb76Bp6cnSpYsiWXLlqFQoUJYuXJlpuvMmzcP69evR506dT64bUNDQ7URvXdvJKJp20RERERERAWBzhK96Oho/Pvvv3B3d38bjIEBvvzyS3h7e+d4+6dOnYKDgwPKly+PgQMHIiQkJM/aJiIiIiIi0iWdJXpvpkg6ODiolTs4OKglZZpwcXHBypUrcefOHezYsQPPnz9HkyZNEBcXl6O2k5KSIJfL1X4KOm9vb9y5c0fXYRARERERkRbp/GYs/322m4GBAXL6xIfJkyejV69ecHR0RN26dbF//34EBQVh165dOWp77ty5sLa2ln5cXFxyFGduOHbsGG7cuCG99vf3x+bNmzOtf+rUKVy5ciWPoiMiIiIiorygs8crvBlNCw8PVysPDw/PMNKWU/b29nB1dcWTJ09y1PbEiRMxduxY6bVcLv9osld9c/UcRq/OZ4DPB5eXLVsWnTp1wtWrV2FpaYk+ffpgypQpWo2BiIiIiIjyN50levb29ihXrhwuXLiAzp07S+Xnz59Hjx49tNpWXFwcXr58KSVxmrZtamoKU1NTrcambRUrVsTIkSMxcuRIlC1bFvXr10erVq0+ut7Ro0cRGBiIbt26wd7eHnfu3MHx48dhYmKC2rVro1mzZkD6VE8LCwuEhIRAoVDAwMAAVatWxc2bN5GSkoLevXvDx8cHFy5cwJdffolq1apJbZw+fRoPHjxA9erVpe0dPHgQ1atXx40bN5CYmAhPT0/Exsbi77//Rt++fQEASqUSv//+O0aNGpVr/UZEREREpE90OnVzzJgxWL9+Pc6fPw+FQoGZM2fi1atXGDZsmFRn5MiR730GXmaSkpLQr18/PHjwAEqlEoGBgejbty9MTU3Ru3fvbLVdUH333XcIDg7GwYMHMXfu3I/WX758Ofbv34/Tp0+jZcuWQHo/RkdHIzg4GOPGjcPvv/8OpE/17Nu3L/bu3YvY2Fhs27YNPXr0gLe3N5YtW4bOnTvjp59+wqNHj9C8eXNp1HTixIn4/vvv8fTpUwwfPhw///wzAGDbtm3o3bs3Lly4gM2bN2Po0KGwt7fHwoULERQUBAD4448/4OPz4ZFMIiIiIiJ6S2cjekhPSCIjI9G5c2fExMSgfPnyOHz4MMqWLSvVUSgUSEhIkF5v374d3377rXQtXZs2bWBoaIhJkyZh0qRJMDU1Rfv27dG3b188ePAA5ubmaNq0KS5fvgxHR8dstV1QxcTEIDg4GEqlElFRUShevPgH63ft2lVKvMqXL4+4uDjUqlULISEhePHiBTw8PLBv3z6MGDECANCxY0cpgTxy5AgmT56Mbt264cGDB2jfvj2ePXsGAwMDxMXF4ebNm2jdujXWrFmDZ8+ewdbWFmFhYahZsyamTZsGABg/fjy6d++OxMREVK1aFTKZDKNHj8aKFSswd+5c/Prrr1i/fn2u9xsRERERkb7QaaIHAFOnTsXUqVORkpICY2PjDMtXrFgBlUolve7Zsyc6deqUoZ6JiYlanZ49e0KpVMLQ0FDjtguqYcOG4fvvv4etrS0GDhyIY8eOfbB+iRIlpP9bWFhAoVBg1KhRCAkJQdWqVZGYmIioqCipTsmSJdXWf5NImpubw8HBQbrJjbm5ORQKBYQQEEJIfWxsbKx2TN+0X6hQISQnJwMA+vbtCzc3N7Rq1QpFihRBxYoVtdAzRERERESfBp0nem9klmj995o4IyMjWFpaZmmbH0rystJ2QbR582YoFAppCuqhQ4ewatWqbE9JjYqKQpEiRWBlZYUXL17kKCaZTAZPT0+0aNECLVu2xNGjR6Xr7zLzZqptjx498Mcff+SofSIiIiKiT43OH69A2qNSqfDy5Uu1aY4rVqxARESE2gjau5o0aYJatWpJrwcOHAhzc3Ns2LABtWvXhq2tLRYsWICBAwe+t37nzp2lO49aW1ujf//+0rLWrVujQoUKQPp1gGPHjpWm2S5evDjD+kifUvtGu3btUKZMGTRp0kQr/UNERERE9KmQiZw+tO4TJpfLYW1tjZiYGFhZWUGhUMDX1xelS5eGmZmZrsNT4+3tDW9vb7WyChUqoEuXLjqL6UM2b96MFStW4Mcff0T37t11HU6+kZ/fY0T0iZluraN2Y3TTLtGnjr/z+cJ/848PyTdTNyl3KRQKREdHq5XFx8frLJ6PiY+Px7Bhw5jkERERERFpgIneJ8LDwwMeHh66DiPL3tzhk4iIiIiIso/X6BEREREREekZJnq5gJc9Um7he4uIiIiIsoJTN7XI2NgYMpkM4eHhKFq0KGQyma5DIj0TGxsLpD9mhIiIiIgoM/y0qEWGhoZwdnZGUFAQ/Pz8dB0O6SkrKysmekRERET0Qfy0qGWWlpYoX748UlJSdB0K6SGZTAYTExNdh0FERERE+RwTvVxgaGgIQ0NDXYdBRERERESfKN6MhYiIiIiISM8w0SMiIiIiItIzTPSIiIiIiIj0DBM9IiIiIiIiPcNEj4iIiIiISM8w0SMiIiIiItIzTPSIiIiIiIj0DBM9IiIiIiIiPcNEj4iIiIiISM8w0SMiIiIiItIzRroOgIiIiIgKkOnWOmgzJu/bJCrgNBrRCwwMxNy5c6XX8+fPh42NDdzc3PDixQttxkdERERERETZpFGi98MPP6By5coAgICAAEyfPh3z589H5cqVMW7cOG3HSERERERERNmg0dTNM2fOYMOGDQCA48ePo23btvj222/RpUsXKQEkIiIiIiIi3dBoRE8mk0EulwMATpw4gS+//BIAYGhoCCGEdiMkIiIiIiKibNFoRO/LL7/EgAED0LhxYxw7dgxLly4FAHh7e6Np06bajpGIiIiIiIiyQaMRvd9//x2urq7w9vbGpk2b4OLiAgDYtWsXpk+fru0YiYiIiIiIKBs0GtGLiYnBxo0bM5Tv2LFDGzERERERERFRDmg0olexYkXtR0JERERERERaoVGi5+TkhICAAO1HQ0RERERERDmmUaL3008/YcSIEQgKCtJ+RERERERERJQjGl2jN23aNERGRsLFxQWFCxeGiYmJ2vKIiAhtxUdERERERETZpFGit3z5cu1HQkRERERERFqhUaLn6emp/UiIiIiIiIhIKzS6Ro+IiIiIiIjyryyP6JmZmQEAFAqF9P/MKBSKnEdGREREREREGslyordv3773/p+IiIiIiIjylywneu3bt3/v/7UhNTUVcXFxsLGxyfI6ycnJiIqKgp2dXYa7fr6hUChgaGgIY2PjDMvkcjkSEhLUyoyNjVGkSBEN9oCIiIiISH8FyVU6addZJ63qB41uxqItQgh4eXlh+fLlUKlUKFq0KJYvX46OHTtmuo6/vz9WrlyJTZs2ISwsDGfPnkWzZs3U6uzevRuLFi3Cw4cPoVQqUbt2bSxbtgxubm5SnQkTJmDTpk1qyWWlSpVw7ty5XNpbIiIiIqKCyWVpnE7aFUt00qxe0DjRu3DhAvbt24eAgACkpqaqLfvzzz+ztI2lS5dizZo1OH/+PGrVqoVff/0V3bp1g4+PDypWrPjedfbs2QNra2scO3ZMLXF7Q6lU4sCBA1i1ahU+++wzJCcnY+TIkWjTpg0eP34MW1tbqW779u05DZWIiIiIiPSORone5s2b8d1336Fbt244dOgQvv32W1y7dg23b99Gr169sryd3377DYMHD0adOnUAAGPHjsXy5cuxevVqLFny/vT9xx9/BAAEBQW9d7mhoSF27dolvS5UqBBmzJiBDRs24Nq1a2jVqpVa/ZiYGFhYWMDISKeDm/plurUO2ozJ+zaJiIiIPhGBP1jqOgTKJo2ymwULFmDXrl1o3749Nm/ejFWrVgEAJk6ciBcvXmRpG+Hh4fDz80OTJk3Uyps2bYpr165pElamfH19AQDFihVTKz948CCOHz+OlJQU1K9fH8uWLcNnn32m1baJiIiIiAo6Zys+la2g0eiIPXv2DB4eHgAAExMTxMfHAwDGjx+PEydOZGkb4eHhAAB7e3u18qJFi+LVq1eahPVeiYmJGD16NL744gu1JK527dq4fPkyYmNjERISAicnJ7Ro0QJhYWGZbispKQlyuVzth4iIiIiIKL/RKNFLTk6WnqXn5OSEBw8eAOl3slQqlVnahkwmA9LvuPmu1NRUGBoaahJWBikpKejRowdiYmKwc+dOtWVDhw5F3bp1IZPJYGdnh/Xr1yMhIeGD1+zNnTsX1tbW0o+Li4tW4iQiIiIiItKmHI/Bdu/eHX369IGXlxc6duyIli1bZmk9R0dHAMgwghYWFiYty4k3Sd6///6Lc+fOoUSJEh+sb25uDkdHR2ma5/tMnDgRMTEx0k9gYGCO4yQiIiIiItI2jRK969evS/+fNWsW+vTpgzt37sDd3R1r167N0jasra1Rs2ZNnDx5UipLTU3F6dOn8fnnn0tlcrkckZGR2YovNTUVnp6euHv3Ls6dO/fekTchhNrrsLAw+Pv7o3Tp0plu19TUFFZWVmo/RERERERE+Y1GN2N5c5dMpD9kfNq0aRo1/r///Q+9e/dGgwYN0LBhQyxatAgqlQrDhw+X6owdOxZXrlzB/fv3gfRr7mJiYqRr/KKiohAaGgpLS0tYWlpCpVKhd+/e8Pb2xpEjR2BiYoLQ0FAgPbksVKgQkpKS4O7ujgkTJqBq1aoICAiAl5cXHB0d0bdvX432hYiIiIiIKL/Q+JkCKpUKf//9Nx4+fAgAqFKlClq2bAkDg6wPEnbr1g1JSUn49ddfMWPGDFSvXh3nzp2Dg4ODVMfa2lrthi2HDx/G999/D6TfRXPEiBFA+o1gxo8fj+joaFy4cAEymSzDg9eXLFmC3r17w9TUFL/88gsWLFiA27dvw9bWFk2bNsWkSZNgba2DRwMQERERERFpkUz8dw5jFjx9+hSdOnXCs2fPULJkSQCAv78/KlSogEOHDqFs2bK5EWu+I5fLYW1tjZiYGE7jfBefo0dEpP90ca4Hz/f5Av/Of5r4O58vZCf/0OgavcGDB6NChQoICgrC06dP8fTpUwQFBaFcuXIYPHiwpnETERERERGRFmg0dfPq1avw9fVVm2Lp4OCAlStXokyZMtqMj4iIiPKhILlKJ+0666RVIqKCR6NEz9XVFYmJiRnKExMT4erqqo24iIiIKB9zWRqnk3bFEp00S0RU4Gg0dfO7775D//79ce/ePQghIITAvXv30K9fP3z33Xfaj5KIiIiIiIiyTKMRvZkzZyIyMhI1a9aEqakphBBITk4GADx+/BgzZsyQ6kZERGgvWiIiIsoXAn+w1HUIRET0ARolesuXL9d+JERERFRgOFtpNCmIiIjyiEaJnqenp/YjISIiIiIiIq3g13FERERERER6hokeERERERGRnmGiR0REREREpGeY6BEREREREekZJnpERERERER6Jst33ZTJZFneqBBC03iIiIiIiIgoh7Kc6J09e1b6/7Vr1zB79myMGjUKdevWBQBcv34dv/32G/73v//lTqRERERERESUJVlO9Jo1ayb9f9KkSdi9ezdat24tlXXq1AlNmjTBjBkz8OOPP2o/UiIiIiIiIsoSjR6Yfu/ePTRq1ChDecOGDeHj46ONuIiIKAtKeR3N8zb95rXL8zaJiIgoezS6GUvRokWxffv2DOXbt2+Hg4ODNuIiIiIiIiIiDWk0ojd37lz07dsXBw4cQN26dSGEwI0bN3DmzBns2LFD+1ESERERERFRlmmU6Hl6eqJy5cpYtmwZzpw5AwCoUqUKbt++jerVq2s7RiIiIiIiIsoGjRI9AKhZsybWr1+v3WiIiIiIiIgox3L0wHR/f3+1xy4QERERERGR7mmU6EVERKBFixYoVaoUvvzyS6m8bdu2uHDhgjbjIyIiIiIiomzSKNEbP348LC0tERwcnKF85syZ2oqNiIiIiIiINKDRNXrHjx/HzZs3UaJECbVyNzc3/PPPP9qKjYiIiIiIiDSgUaInl8tRuHBhAIBMJpPKo6OjYWxsrL3oiIjog1LlEboOgYg+MUFyVZ636ZznLRIVfBolevXr18euXbswdOhQKdFTqVSYMWMGmjRpou0YiYgoEy9Xfp33jf4+IO/bJKJ8w2VpXJ63KZbkeZNEBZ5Gid78+fPh4eGBM2fOQAiBH374ASdPnoSfnx+8vb21HyURERERERFlmUaJXr169XD9+nUsWbIEtWrVwrlz51C/fn3s27cPlSpV0n6URET0Xk7DN+k6BCL6xAT+YKnrEIgoCzRK9MaPH49FixZh9erVmS4jIqLcZ2Rlr+sQiOgT42yVo8cwE1Ee0eg3dfHixRotIyIiIiIiotyn1a9kbt26BQcHB21ukoiIiIiIiLIpW1M3LS0t3/t/pN91MzExEePHj9dedERERFkQFBSU5206O/OG70RElH9lK9Hbtm0bAKBz587S/98wNjZGqVKlULVqVe1GSERE9BEuLi553qYQIs/bJCIiyqpsJXpfffUVAODixYt8Xh4REREREVE+pdFdN5nkERFRfhIYGKjrEIiIiPIVjRI9ALhw4QL27duHgIAApKamqi37888/tREbERUk06111G6MbtqlfIXXyxEREanT6K6bmzdvRtu2bSGXy3Ho0CE4OzsjODgYR48ehZWVlfajJCIiIiIioizTKNFbsGABdu3ahU2bNgEAVq1ahVu3bsHLywtKpVLbMRIREREREVE2aJToPXv2DB4eHgAAExMTxMfHAwDGjx+PEydOZHt7gYGBuHHjBuRyeZbXCQkJgbe3N2JiMp+2lZXtatI2ERERERFRfqZRopecnAwzMzMAgJOTEx48eAAAkMvl2RrRUygU6Nq1KypWrIh+/fqhePHi+O233z64zo0bN9CtWzfUrFkTTZs2xe3btzXariZtExERERERFQQa34zlje7du6NPnz7o0qULjh49ipYtW2Z53Z9//hnXrl3D8+fPUaJECRw8eBCdO3dGvXr1UL9+/feuc//+ffTs2RMLFy5EmTJlNN6uJm0TEVFGfFg5ERFR/qNRonf9+nXp/7NmzYK5uTkuX74Md3d3TJs2Lcvb2bhxI4YPH44SJUoA6c/pq1atGjZu3JhpsvX1118DH/lgkZXtatI2ERFlxIeVExER5T8aTd3ctWuX9H9jY2NMmzYNx48fxy+//ILZs2dnaRvBwcEICwuDm5ubWnm9evXeOx0zq7KyXU3bTkpKglwuV/sBgKRHL6Q6yY99kfoyDACgUiQh6e5jqOISAACpr6KQdP/Z27rPApASGAoAECmpSLr7GEp5HABAGfEaSfeevK37PBAp/sFpdZXKtLrRsWl1o2KQdPex9MEnxfclUnxfptUVIq1uVNq1jMro2LS66VNsU/yDkfz87fOnku49gTLidVpdeVxa3ZS0x2ekBIYi+VnA27r3nyH1VVTavsYlpO2rIiltX2ONkRxl+rZupBlS49O+V1ClyJAUXgiqVFla3TgjJEe+rZscaYrUOOO0uqnpdZPT3qqp8UZIijB7WzfKFCmxaXVFUnJaDLHxb/vb56l6fweEqPd3TOzb/r77WL2//f7T36/TjrfytTytX1SqtH7xC0bKi7dfPCTdfQxlZLR6f6c/giStv9/pQ5+nSA1P629VbHxa3eSUtLpBYUh+6v+27r/PkBoW+f7+Dn6F5Me+b+s+fIHU0Ii0ugmKtLoJirS6oRFIevif92zwq7S6/33PhkUi6d933rNP/ZESlPb+Fskp6v2dYIik8HeOTbQJUuTpx0YFJIUXglJhmNYviYZICi8k1U2JNkFKjElaXfGfuoq0ukKVXjfGBMnRJur9/eY9GxOr/p4NCFF/z/o8ffuefdPfSclv+/uJn3p/v+nDN/2d+E5/P3rb38mPfJEaEp5WNzG9D+MT3/b3g+fq/Z0L5winmCgUj0173xmoVCgfEQrLpLRjbqVIRPmI0LTOBVBC/hol5Gl9BiFQPiIUVoq0eC2TFCgfEQqD9Pd38dhoOMVESe2UiwjNcI4wQtrvsqOBKUobvj2ulQwtUESW9h6wkBmiqqEFTNLrljAwQbl36lYwNEfR9LqFYICqhhYwTf8TVczABOUNzaW65Q3N375nE//Th6ER6u/ZJ35v37M8R7ytq61zxH/Pya/fOScr/3P+1tY5QqV6f3+/+bvm95+/azxHpNXV5ueId8/JcmP1c3K4GZSJ6ccxySCtrvKduq/f+WwQYYbUhPTPBskG6p8N/vs54sHzt/0dn6je3yHhGfub5wjtnyNS0o9NvBGS3v3cFpWNz22aniNSU9Xfs5HRav2d8iLobX/r+zni6ds2PkajRG/x4sUaLXtXVFRaJxUpUkStvEiRItIyTWRlu5q2PXfuXFhbW0s/b77FDv1milQnbOh0RC/fCQBQBocjyGOw9EaM3XMcwZ1HS3VfjZqD14vT7lyqjIxGkMdgKK7cAwDEHTqLoDbDpLoRExYjavYaAIBIUCDIYzASz98AAMT//Q+CPAYD6W/giCm/IWJK+vWGSiWCPAYj/u9/AACJ528gyGMwRPof9KjZaxAx4e0xC2ozDHGHzgIAFFfuIchjsHSyeb14E16NmiPVDe48GrF7jgPpvyRBHoOhDE57k0bfLoqwE65S3ZAjpSD/N62/k6PMELSnPFKi004SMT72CDlWSqoberwUou/aAwBSY0wQtKc8kqPSfvFjH9oh5Ejpt/190hXRt4qm1Q2LTOvDWw/T+nD/CbzsOFKqG/79PLxesAEAoIqJTat7+W5a3aMXENTq27f97bUUkTNWpvW3IhlBHoORcPYaACDh1OW0/k4/CUROX4GIyb++7UOPwYg/djGtv71vIchjMFSxaX9gouatQ/jYhVLdl+1HIO7A6bT+vn4/rQ/TTzbRS7fg1YhZb/u76xjE7jyW1t/3nyHIYzBS0/84Ra/cjdBBU9/2YZ+fELPhAAAg5ak/gjwGIyX9ZB+z4QBC+/z0tu6gqYheuTutD4PC0t6z6R8kYnceQ3DXMVLdVyNmIXrpFiD9j1qQx2Aort9P68OnNnj5R7m3/X3WGVFXi6f1d7IhgvaUR+JLCwBAvK81gvaUf9vf3o6IvJQ2ug6VDEF7yiPBv3BafwcURtCe8hDKtD8wkZeLI+KC09v+bvUt4o5eSOvDy3fT+jv9D+/rBRsQ/v28t/3dcSTi9qfdMEpx62FaH6b/0Yteth1hw2ZIdUN6jEPs9qMAgOSHL9Lqpn9Iil6zF6FfT35bt99ExKzbn9bfLwIR5DFYOtnLtxxGiOePUt3cOkd8f+lvDLpxHgBglpqCVYc2oXZwWgwNA55i1aFNMEj/EDfi6mmMuJr2vjMQAqsObULDgLQPM7WD/bDq0CaYpaZ9UBh04zy+v/S31M5vR7ZmOEe8uHUXgYGB8B44Aac+90RgYCACAwPxZ8kmuDHzVwQGBuLBH8dwyLYWnv1zDYGBgbg0Ygr+rveVVPdYRQ9cm7QAgYGBeHT8DA7Z1sKT0xcQGBiIK+Nm4a8abaS6xz9rh+g1e9Pes/7Baf2dnpjEbj+KkB7j3vb3sBmIXrY9rS7PEVJdrZ0jHtki+NDbyyhenXJB9A2HtPdsohGC9pSHIjQtSdfaOSIlNa2/T11O6++z19L+rinSPmxFzliJCK+lb/ub54i0PtTm54g95SFS0z5CRl0pgYjz75yT95VD3DObtD4MsUDQnvJQKtKSudc3iuHV6bfTroMPlkHsI1sAaYl+0J7yUKYnDBk+R3j+CPmWw2n9/cQv7T37Iu3Desy6/QjpN1GqG/r1ZJ4jcuMcEZ32WSzGpwhCj779LBZ6vCSi76R/botN+9z2JrnT2jkiNiHtPet9CwAQf+xiWh++6e/JvyJy+oq0F3p+jng1ei6ySiY0mP8ik8neO23m1q1baNOmDcLCwj66jcePH6NSpUo4c+YMmjdvLpWPGjUKZ86cwb///vvB9YOCguDi4oKzZ8+iWbNm2dqupm0nJSUhKSlJei2Xy+Hi4oJXV2+jaL3PgPRv4gwszWHkVAwqRRJSHvvBuKwLDCzNkfoqCspXUTCtlvYmTn4WAJmpCYxdikOkpCL5wXMYlXaCoZUllBGvkRocDtMaFdLqPg+EzMgQxiUdIZRKJN9/BqOSjjC0KQxlVAxSA0NhUqMCZDKZ9C2ccWknCCGQfO8JjFyKw9DOGsroWKT6B8OkWjnIDA2R4h8MkaqESdm0pDXp3hMYORaFob0tlPI4pPq+hEmVspAZGyElMBQiKRkm5dJOvEn3n8HQwQ5GDnZQxSUg5XkgjCuWgsE8B6TGGkOVYgATu7T+Soo0g6FZKowsUqFKkSEl2gzGtgoYGAmkxhlBlWQIkyJpdZMjTWFgqoKRZQpUqTKkvDaDsXUSDExUSI03gjLRCKb2aYlqcpQpZMYqGC+OgEhKRvIjXxiXcYZBYYu0/g6LhGn18m/728QYxq4l3vZ3KUcYWhdO6++Xr2Bas+Lb/jY0hHGpd/rbtQQMba2gfC1HakAITKqXh8zAIO0bJJUKxmXS/ngl3X0MI+diMCxi87a/q5aFzMgovb9TYVI2vQ99nsKwuD2MitpCFRuPlBdBMKlcBjITY6QEhUEkKmBSvmRa3X+fwdDeFkbFiqj3t5kpUoNfQRUbD5OKaSfepIcvYGhrBaPi9lAlKJDy1B/G5UvCwNwMqaERUL6Ww7Rymbfv2cIWMHJ0yPieDYuEMuI1TKumv2ef+kNWyAzGzsUgklOQ/PDF2/6eYAdlvDFMi6Yfm2gTyAwEjK1SIFRAcmQhGBVOhqGZEspEQ6TGmcC0aNq3VSnRJoAMMLZOhhBAcsQ7dRWGSI01gUmRRMgM0kb0hABMbNJO2Emdr8HIySHtPRsTi1S/4Lfv2YAQiOSUt+9Zn6cwLFYk7T37pr8rlYbM1CStvxMSYVKh1Nv+LmKT1odv+rtCKRgUSu9veTxMKqX1d/IjXxhYW8KoRFGoEpOQ8sQPxuVcYWBRKK2/o2JgWqVsrpwjKu5IS9CcYqKgNDBAaGEbGKhUKBv1CiGFbRBnagYrRSKKxcXgaZFigEwmjeaFWNmmjehFhiHM0hpys0KwTFKgRGw0nts5QGVggOKx0TBUqfDS2g5IH9H7e16XnJ0jzEyR+jIMqriEt+/ZB89haGed1t/xiUh5FvC2v0PCoYqJU+9vq/T37Jv+ftOHoRFQRka/fc8+8YPMvFDae5bnCO2eIxaVyHhOfm0KmZEKxoVTIJQyJEe9c/5OMNTOOWLqayT7PM3Y32/+rvkFQyjf+bt29/EnfY7Ilc8Rk8u/PSfLjSFUsrfn5HAzGFmmwLCQEsokA6TKTWFilwiZYXpdpQFMbNM/G0SYwdA8FUbmqVAlGyAlxvTtZ4P/fo7ocYvnCF2fI/bUh4GxSPu9VxjC9M3ntihTGJhk8XObJueIJREQqalI/vf52/dsZDRSg8Kk/k55EQQYGKT1t0ql1+eIiJv3ULROTcTExHz0+eXZSvQsLS0BAPHx8bCwsFBbplKpkJiYiPHjx2PhwoWZbOGthIQEFC5cGJs3b0bfvn2l8i5duiAxMRF//fXXB9fPLNHLynZz2vYbcrkc1tbWWeroT8p0ax20mfljNiiP6OK4g8e+lNfRPG/Tb167PG+T8iH+zn+6+Hf+08Tf+XwhO/lHtqZubtu2Ddu2bVP7/5ufvXv34v79+1lK8gDA3NwcjRo1wuHDh6Wy+Ph4nDp1Ci1atJDKnj17lq1r9rKy3ay2TUREREREVBBl666bX331FQDg4sWLaNKkSY4bnzVrFlq0aIGJEyeiYcOG+O233+Dg4IChQ4dKdebNm4crV67g/v2064BevXqFJ0+eIDw87VowHx8fGBkZwdXVFa6urlneblbqEBERERERFUQa3YzlTZKnUqkQGBiIgIAAqNLvHJQdX3zxBc6ePQt/f3/8+uuvqFq1Kry9vaUpogBQvnx51K5dW3p98+ZNeHl5YfHixWjcuDF2794NLy8vnDp1KlvbzUodIiIiIiKigkij5+gplUrMmTMHCxcuRGxs2l1pChcujB9//BGTJk2CoaFhlrfVuHFjNG7cONPlP/30k9rrNm3aoE2bNjneblbrEBERERERFTQaJXrTpk3D2rVrMWfOHDRo0AAymQyXL1/GzJkzkZSUhFmzZmVhK0RERERERJQbNEr0NmzYgL179+Lzzz+Xytzc3FCjRg306tWLiR4REREREZEOaXSNXkREBGrWrJmhvEaNGoiIiNBGXERERERERKQhjRK9KlWqYMOGDRnK169fjypVqmgjLiIiIiIiItKQRlM358yZg06dOuHQoUOoV68eAODq1au4dOkSDh06pO0YiYiIiIiIKBs0SvTatm2Lu3fvYsmSJbh48SJkMhmqVKmClStXckSPiIjyXPXN1fO8TZ8BPnneJhERUVZplOghffrmunXrtBsNERERERER5ZjGiR4ApKSkICYmJkO5vb19TjZLREREREREOaBRovfgwQMMGjQI165dg0qlyrBcCKGN2IiIiIiIiEgDGiV6Q4YMQbFixXDq1CnY2tpqPyoq0ILkGZP/3Oac5y0SEREREeVfGiV6d+7cQWBgIOzs7LQfERV4Lkvj8rxNsSTPmyQiIiIiyrc0eo6eq6srEhIStB8NERERERER5ZhGI3rff/89vv/+e6xdu5ajepRB4A+Wug6BiIiIiOiTplGi97///Q+RkZH4448/YGdnB5lMprY8IiJCW/FRAeRspdFAMRVwurg2E7w+k4iIiOi9NEr0li9frv1IiKhA08W1meD1mURERETvpVGi5+npqf1IiIiIiIgog1JeR3XSrt+8djppl7SDD0wnIq3gtZlERERE+QcfmE5EWsFrM4mIiIjyDz4wnYiIiIiISM/wgelERERERER6RqNE780D05noERERplvroM2M14cTERHRWxpdVPPmgelRUVHaj4iIiIiIiIhyhA9MJyIiIiIi0jN8YDoREREREZGe4QPTiYiIiIiI9AwffEVERERERKRnNBrRA4ALFy5g3759CAgIQGpqqtqyP//8UxuxERERERERkQY0GtHbvHkz2rZtC7lcjkOHDsHZ2RnBwcE4evQorKystB8lERERERERZZlGid6CBQuwa9cubNq0CQCwatUq3Lp1C15eXlAqldqOkYiIiIiIiLJBo6mbz549g4eHBwDAxMQE8fHxsLCwwPjx41GuXDltx0hERERERETZoFGil5ycDDMzMwCAk5MTHjx4gLp160Iul3NEj/KFoKCgPG/T2dk5z9skIiIiInofjW/G8kb37t3Rp08fdOnSBUePHkXLli21ExlRDri4uOR5m0KIPG+TiIiIiOh9NEr0rl+/Lv1/1qxZMDc3x+XLl+Hu7o5p06ZpMz4iIiIiIiLKJo0SvTp16kj/NzY2ZnJH+U5gYKCuQyAiIiIi0pkcT90kyo94vRwRERERfcqY6BERERERUb5XfXN1nbTrM8BHJ+3mlEbP0SMiIiIiIqL8K8uJ3rBhw6T/nzt3LrfiISIiIiIiohzK8tTNtWvX4vfff4eBgQGaN2+utVvJp6Sk4Ny5cwgLC0P16tVRs2bNHK+zefNmpKSkZFivUqVKaNKkCQDgn3/+wcOHD9WW29nZoUuXLjneJyIiIiIiIl3KcqJXsmRJrFy5Eo0aNQIA3LlzJ9O6n332WZa2GRkZCXd3d8TFxaF69er47rvv0K9fPyxfvjxH61y/fh0KhUJ6HRsbiz179mD27NlSord161b8/fffcHd3l+o5OTkx0SMiIiIiogIvy4neokWLMHToUERGRgIAatWqlWndrI72TZw4ESkpKbh79y4sLCxw/fp11K9fH+3atUObNm00Xue/ieKqVauwf/9+fP3112rlbm5uWLduXZZiJSIiIiIiKiiyfI1ely5dEBERgdevXwMAwsPDM/3JCpVKhd27d2PgwIGwsLAAANStWxcNGjTAzp07tbYOAKxfvx7t2rWDo6OjWnl4eDh27NiBo0ePIiwsLKtdQURERERElK9l+/EKNjY2+Ouvv2Bvb5+jhgMDAyGXy1GlShW18qpVq+LmzZtaW+fevXu4ceMGjhw5kmHZ8+fPcfDgQbx8+RK3b9/GokWLMGLEiExjTkpKQlJSkvRaLpd/dD+JiIiIiIjymkaPV2jdujWQPsIWGBiIgIAAqFSqbG3jTZJka2urVm5nZ5dpAqXJOuvWrYOTk1OGqaADBgyAr68v9uzZg3/++Qe//vorRo8ejVu3bmUa89y5c2FtbS39uLi4ZHFviYiIiIiI8o5GiZ5SqcTMmTNhY2MDV1dXlCxZEjY2Npg5cyaUSmWWtlGoUCEg/UYp74qNjZWW5XSdpKQkbN++HQMHDoShoaHasoYNG8LY2Fh6PWTIENjZ2eHEiROZxjxx4kTExMRIP4GBgVnaVyIiIiIioryU7ambADBt2jSsXbsWc+bMQYMGDSCTyXD58mXMnDkTSUlJmDVr1ke34erqCmNjY/j5+amV+/r6oly5clpZ58CBA4iOjsbAgQOztF+FChVCVFRUpstNTU1hamqapW0RERERERHpikYjehs2bMDevXsxcuRI1KlTB25ubhg5ciT27t2LjRs3ZmkbJiYmaNWqFXbt2iXdpTMkJARnz55Fhw4dpHre3t74448/srXOG+vXr0eLFi1QqlQptXKlUplhNO7y5csIDAxEw4YNNegRIiIiIiKi/EOjEb2IiIj3Pti8Ro0aiIiIyPJ25s+fj0aNGqFr165o0KABNm/eDDc3N/Tr10+qs2nTJly5ckV6vl1W1gEAPz8/nD59Gnv37s3QrhACbdq0QaNGjVC1alUEBARgzZo16NGjBzp16pTN3iAiIiIiIspfNBrRq1KlCjZs2JChfP369RnuiPmx7dy7dw+fffYZ/P39MWrUKJw5c0bt2rmmTZuia9eu2VoH6XfUHDp0KDp27JihXSMjI9y8eRNNmjSBn58frK2tcfjwYezatQsGBhp1CRERERERUb6h0YjenDlz0KlTJxw6dAj16tUDAFy9ehWXLl3CoUOHsrUtV1dXTJ06NdPlAwYMyPY6AODu7g53d/dMl5uamqJ///7o379/tuKlgqH65up53qbPAJ88b5OIiIiI6H00Gr5q27Yt7t69i3LlyuHixYvw9vZG+fLlcffuXbRt21b7URIREREREVGWaTSih/QplOvWrdNuNERERERERJRjvCCNiIiIiIhIzzDRIyIiIiIi0jMaT90kIiIiorxVyutonrfpN69dnrdJRDnHET0iIiIiIiI9w0SPiIiIiIhIz2R56qaZmVmWN6pQKDSNh4iIiIiIiHIoy4nevn37cjcSIiIiIiIi0oosJ3rt27fP3UiIiIiI6INS5RG6DoGICgjedVPPBAUF6aRdZ2dnnbRLRET0KXm58uu8b/T3AXnfJhHlGK/R0zMuLi46aVcIoZN2iYiIiIgoI16jR0RERFRAOA3fpOsQiKiA4DV6eiYwMFDXIRAREVEuMbKy13UIRFRA8Bo9PcNr5YiIiIiISONE78KFC9i3bx8CAgKQmpqqtuzPP//URmxERERERBno4uZz/DKdChqNEr3Nmzfju+++Q7du3XDo0CF8++23uHbtGm7fvo1evXppP0oiIiIionS6uPkcbzxHBY2BJistWLAAu3btwqZNaRcEr1q1Crdu3YKXlxeUSqW2YyQiIiIiIqJs0GhE79mzZ/Dw8AAAmJiYID4+HhYWFhg/fjzKlSun7RiJiIiIMuCzYz9dvPkc0cdplOglJydLz9VzcnLCgwcPULduXcjlco7oERERUZ7gs2M/XUy2iT4ux3fd7N69O/r06YMuXbrg6NGjaNmypXYiIyIiIiIiIo1olOhdv35d+v+sWbNgbm6Oy5cvw93dHdOmTdNmfERERETvxel7RESZ0yjRq1OnjvR/Y2NjJndERESU5zh9j4gocxrddTM1NRU3b97MUH7z5s0Mz9QjIiIiIiKivKXRiN6cOXOgVCrh5uamVn748GEcO3YMU6ZM0VZ8RERERERqqm+unudt+gzwyfM2iXJCoxG9NWvWYNiwYRnKv/32W6xfv14bcREREREREZGGNEr0Xr9+nemyV69e5SQeIiIiIiIiyiGNpm42aNAAixcvxqJFi9TKFy1ahAYNGmgrNiIiInoPPiiciIg+RuNr9Jo3b46LFy+iadOmEELg4sWLuH//Ps6ePav9KImIiEjCB4UTEdHHaJTo1a9fHzdu3MDixYtx+vRpyGQy1KpVC5s2bUKVKlW0HyVlWSmvozpp129eO520S0REny5d3JADvCkHERUQGiV6AFClShXeeIWIiEgH+KBwIiL6GI0TPQDw9/fHixcv0Lx5c+1FRERERB/Ea+WIiOhjNEr0IiIi0KtXL5w6dQp4Z85+27Zt4eXlhc8//1y7URIRERER0SctJSpF1yEUKBoleuPHj4elpSWCg4Ph6OioVj5z5kycPHlSmzESEREREdEn7vHYx7pp+AfdNJtTGiV6x48fx82bN1GiRAm1cjc3N/zzzz/aio2IKFsqLa2kk3Yf/fBIJ+0SERERZUajRE8ul6Nw4cIAAJlMJpVHR0fD2NhYe9EREWUDv+kjIiLSXxWXVNR1CAWKxo9X2LVrF4YOHSoleiqVCjNmzECTJk20HSMREREREX3ijO04oJQdGiV68+fPh4eHB86cOQMhBH744QecPHkSfn5+8Pb21n6URERZwG/6iIiIiNJolOjVq1cP169fx5IlS1CrVi2cO3cO9evXx759+1CpUvaukYmMjMSePXsQFhaG6tWro3PnzjAwMMjROgcPHsSVK1fU1ilevDjGjBmT47aJKP/iN31EREREaTRK9BYtWoTx48dj9erVOWrc19cXjRs3RoUKFVC3bl2MGzcOGzZswJEjRzJNuLKyzvHjx/HPP/+gT58+0npvrinMSdtE+VUpr6M6addvXjudtEtERPQpSZVH6DoEKoA0SvQmT56MMWPGwMgoR89bx08//YRSpUrh9OnTMDQ0xPDhw1GxYkXs3r0bvXr1ytE6FStWhJeXl1bbJiIiIiLKay9Xfq2bhn8foJt2SSs0GrqqUaMGrl69mqOGU1NTceTIEfTt2xeGhoYAgDJlyuCLL77AH3/8keN1/Pz8MGPGDPz666+4efNmjtsmIiIiIiIqKDQakvP09ETPnj3h5eWFKlWqwMTERG15Vu68GRAQAIVCgXLlyqmVlytXDpcvX87xOqamplAoFHj+/DkmTJiAUaNGYdGiRRq3DQBJSUlISkqSXsvl8o/uJxERERFRTjgN36TrEKgA0ijRGz9+PABg1KhR710uhPjoNuLj4wEAVlZWauXW1tbSMk3XGT9+vFoS17t3b7Ru3Rrt27dHs2bNNGobAObOnYuff/75o/tGRERERKQtRlb2ug6BCiCNEr3ExMQcN2xpaQkAiImJUSuPjo6Wlmm6zn9H6lq1agUnJydcuHABzZo106htAJg4cSLGjh0rvZbL5XBxcfnovhLlNl6kTURERETv0ijRMzMzy3HDrq6usLCwwOPHj9GqVSup/PHjx6hcubLW1nlDqVRCoVDkaDumpqYwNTXN1n4S5QVepE1ERERE79LZcwQMDQ3RuXNnbNmyBcnJyQCABw8ewNvbG927d5fq/fHHH/jll1+yvE5qamqGZ+jt378foaGhcHd3z1bbREREREREBVHOno+QQ/PmzUPTpk3RsGFDuLm54fDhw+jWrRs6d+4s1Tl27BiuXLkiPez8Y+vIZDJ4eXnBwMAAVatWRUBAAP7++29MnjxZSvSy2jZRQcGLtImIiIjoXTpN9JycnHDv3j0cPnwYYWFh6NGjBzw8PNTqdO3aFQ0bNszyOoaGhjh37hwuXbqE27dvo2HDhli6dCnKlCmT7baJCgpepE1ERERE79Jpoof0G6z07t070+Vt2rTJ9joA0KhRIzRq1ChHbRMRERERERVEOrtGj4iIiIiIiHIHEz0iIiIiIiI9w0SPiIiIiIhIzzDRIyIiIiIi0jNM9IiIiIiIiPQMEz0iIiIiIiI9w0SPiIiIiIhIzzDRIyIiIiIi0jNM9IiIiIiIiPSMka4DICIiouwp5XVUJ+36zWunk3aJiCj7OKJHRERERESkZ5joERERERER6RkmekRERERERHqGiR4REREREZGeYaJHRERERESkZ5joERERERER6RkmekRERERERHqGiR4REREREZGeYaJHRERERESkZ5joERERERER6RkmekRERERERHqGiR4REREREZGeYaJHRERERESkZ5joERERERER6RkmekRERERERHqGiR4REREREZGeYaJHRERERESkZ5joERERERER6RkmekRERERERHqGiR4REREREZGeYaJHRERERESkZ5joERERERER6RkmekRERERERHqGiR4REREREZGeYaJHRERERESkZ5joERERERER6RkmekRERERERHqGiR4REREREZGeMdJ1AI8ePcLatWsRFhaG6tWr47vvvoOlpWWO1klKSsKuXbtw6dIlGBkZoUmTJujZsycMDN7mtWvWrMGJEyfUtuvq6oolS5bkwl4SERERERHlHZ2O6N26dQtubm6IiopC06ZNsW/fPjRt2hRJSUkar6NSqVClShWcO3cObm5uKF++PH788Ud06dIFQgi17bx8+RKenp7ST5s2bfJkv4mIiIiIiHKTTkf0Jk6ciGbNmmHjxo0AgG7dusHFxQUbN27EsGHDNFpHJpPhwoULcHJyktapX78+GjVqhJs3b6JOnTpSuZOTE7p165br+0lERERERJSXdDail5SUhDNnzqglWkWKFIG7uzuOHTum8ToymUwtyQMAZ2dnAMDr16/Vyu/fv4/+/ftj1KhR2LNnj1b3j4iIiIiISFd0lugFBAQgNTUVrq6uauWurq548eKF1tYBgN9++w02NjaoX7++VGZkZIT69eujWbNmcHBwwIgRI9C5c+cPxpyUlAS5XK72Q0RERERElN/obOrmm2vqzM3N1cotLS2hUCi0ts6ePXuwePFi7Ny5E1ZWVlL5zJkzYWtrK71u37493NzccOjQIXTq1Om925o7dy5+/vnnLO8jERERERGRLuhsRM/a2hp4z3TKyMhI2NjYaGWdQ4cOoV+/flixYgV69OihtuzdJA8AatWqhZIlS+L69euZxjxx4kTExMRIP4GBgR/dTyIiIiIioryms0TP2dkZtra2uHfvnlr5vXv3UKNGjRyvc+TIEfTs2RNLly7N9MYu7xJCIDY2FoaGhpnWMTU1hZWVldoPERERERFRfqOzRE8mk6FPnz5Yv349oqOjAQDnz5/H9evX0bdvX6neqlWrMHbs2Gytc/ToUXTv3h1LlizBiBEjMrSdkpKC7du3q5UtXrwYUVFR6NChQ67tMxERERERUV7Q6eMVZs+ejVu3bqFy5cqoXLkyrl69ikmTJuHLL7+U6ty4cQNXrlzJ8jpyuRxdu3aFlZUVzpw5gzNnzkjrDh8+HO7u7jA0NMSpU6cwZcoUVKxYEYGBgQgJCcH69evVHr9ARERERERUEOk00bOysoK3tzeuX7+OsLAwVKtWDaVLl1arM3z4cPTs2TPL65iZmWHbtm3vba9s2bIAAAMDA2zcuBEhISG4d+8ebG1tUbVqVVhYWOTavhIREREREeUVnSZ6SJ+OWa9evUyXu7m5ZWsdExOTLD8EvUSJEihRokQ2oiUiIiIiIsr/dHaNHhEREREREeUOJnpERERERER6hokeERERERGRnmGiR0REREREpGeY6BEREREREekZJnpERERERER6hokeERERERGRnmGiR0REREREpGeY6BEREREREekZJnpERERERER6hokeERERERGRnmGiR0REREREpGeY6BEREREREekZJnpERERERER6hokeERERERGRnmGiR0REREREpGeY6BEREREREekZJnpERERERER6hokeERERERGRnmGiR0REREREpGeY6BEREREREekZJnpERERERER6hokeERERERGRnmGiR0REREREpGeY6BEREREREekZJnpERERERER6hokeERERERGRnmGiR0REREREpGeY6BEREREREekZJnpERERERER6hokeERERERGRnmGiR0REREREpGeY6BEREREREekZJnpERERERER6hokeERERERGRnmGiR0REREREpGeY6BEREREREekZJnpERERERER6RueJ3qlTp9CpUyc0aNAAQ4YMQWBgoFbW0VYdIiIiIiKigkanid7JkyfRpk0b1K1bF7Nnz0ZYWBgaN26MmJiYHK2jrTpEREREREQFkU4TvalTp6Jnz5743//+B3d3d+zZswexsbFYvXp1jtbRVh0iIiIiIqKCSGeJXlxcHK5evYq2bdtKZWZmZnB3d8fp06c1XkdbdYiIiIiIiAoqI101/PLlSwghUKJECbVyR0dHnDp1SuN1tFXnfZKSkpCUlCS9fjPNUy6XZ3Gvc58qKUEn7ar1QZLQRQBqL5WJSh2EoLv3wSd73KF+7HVx3PEJHvsM+8vf+TzH3/k0/J3PG/ydfxPCp3Xcwd/5d0LIP5/138QixMePh84SvZSUFACAqampWrmpqam0TJN1tFXnfebOnYuff/45Q7mLi0um63wqrH/RcQDzrHUcAGA9XPcx5DWdH3fw2OsCj3uaT+24g8de8qkdex73NJ/acQePvSQ/HvvY2FhYW384Lp0lekWKFAEAREZGqpVHRkZKyzRZR1t13mfixIkYO3as9FqlUiEqKgpFihSBTCb76D7Th8nlcri4uCAwMBBWVla6DofyCI/7p4vH/tPE4/5p4nH/dPHYa5cQArGxsXB0dPxoXZ0leiVKlICjoyOuXbuGDh06SOVXrlxB8+bNNV5HW3Xex9TUNMMooI2NjUb7T5mzsrLiieATxOP+6eKx/zTxuH+aeNw/XTz22vOxkbw3dHrXzSFDhmDdunXw8/MDAOzYsQOPHj3CoEGDpDozZsxAt27dsrWOtuoQEREREREVRDob0QOAyZMn48WLF6hYsSKKFSuG6OhorF27FrVr15bqBAQE4NGjR9laR1t1iIiIiIiICiKZyMotW3JZREQEwsPDUapUKRQqVEhtWWBgIBISElCxYsUsr6PtOpQ3kpKSMHfuXEycODHDFFnSXzzuny4e+08Tj/unicf908Vjrzv5ItEjIiIiIiIi7dHpNXpERERERESkfUz0iIiIiIiI9AwTPSIiIiIiIj3DRI+IiIhyxbBhw3Dw4EFdh0FE9EliokdEuS45ORkXL17EqVOnkJKSoutwiCgXrV69GnPnzgUANGjQAFu2bNF1SEREnyTedZOIctXz58/RunVr2NvbIzw8HEWLFoW3tzcMDQ11HRoR5YKbN2/C09MTT58+RWxsLFxdXfH8+XPY2dnpOjQiymWvX7+GlZUV/8bnExzRozz3559/Yty4cbh165auQ6E8MGjQICxbtgyXL1/Gw4cPERISgmvXruk6LMojKpUKa9euhZeXF4KCgnQdDuUBNzc3mJiY4MqVKyhcuDDatGmDPXv26DosyiMqlQrr16+Hl5cXAgMDdR0O5ZFr167Bzc0NFSpUgLOzM44eParrkIiJHuUFlUqFq1evIiQkBKtWrcLixYshk8nQokULhIaG6jo8ymWrVq1CmzZtAADGxsawtbWFmZmZrsOiXJSYmIh//vkHcrkc3377Lc6cOYPQ0FC0atUKSqVS1+FRHujXrx+2bt0KAOjfvz+nb+o5hUIh/c6PGDECJ0+exKtXr9CiRQukpqbqOjzKBUqlEps2bUJgYCBev36NAQMGYN68ebh16xY6d+6MIUOG8NjnA5y6SbnGx8cHQggMHz4cERERCA0NhZubG06dOgUDAwOMHj0aVlZWmDVrlq5DJS179eoVpkyZgtWrV6uVP3/+HJ06dYKPjw9kMpnO4qPccerUKdjZ2cHT0xMGBgaIjY1F69atsX79egBAo0aNMGHCBHz11Ve6DpW0KCEhAbt370ZgYCA6dOiAWrVqITAwEG5ubggKCoKhoSGcnZ1x8eJFlCtXTtfhkhadOXMGNjY26NWrF2QyGeRyOVq2bIlNmzYBAD7//HOMHj0a3bp103WopCWRkZFYsGAB4uLi4Ovri19++QV+fn6YNm0aLCwscOPGDXTu3Bn9+vVD8+bN+bdexziiR7nm+PHjaNGiBQYNGoTHjx9j8eLFSEhIgIFB2ttu4MCB2LZtG/hdQ8GXmpqKyMhI6XV4ePh7p2cuXLgQP/74I0/8BYivry9GjBiRpbpr165Fp06dsHPnTjx69AgdOnRAoUKFpOUDBw6URnlIPwQFBaF69eo4ceIEjIyM0LVrV9y+fRsuLi6oVq0ajh07BkNDQ/Tq1Qvbtm3TdbikZevXr0fHjh2xdetWPHr0CF26dFGbscHfef0ghMCPP/6I1NRU2NraYtu2bQgJCcGxY8dQoUIFFCtWDDdv3sQ333yDkJAQbNy4EV9++SX/1ucDTPRIK44ePQpPT094enpKH/D79OmDiIgIdOrUCUifyvP06VNpzv5nn30GKysrXLx4UaexU84dOnQIDRs2lJI9pVKZ4ULsly9f4tSpU+jduzcSExNx5MgRHUVL2eHs7Iz9+/erXV+nVCqxevVqdOrUCUOGDJGW9evXD0ZGRnBzcwMAjBo1CgcPHoRKpQIAdO/eHadPn8br1691tDf0IfHx8R+tc/XqVQwePBg7duwAAMycORMzZ85Ejx49cP36dUREROD8+fPAf6Zv9uvXj4lePiWEwJ49ez74pasQAlu3boWHhwc+//xz6fzdr18/yGQy1KtXD0j/nT906JA0Rbtbt244f/48IiIi8mhvSFtevHiBxMREAIBMJsONGzdw/PhxGBgYoHfv3nBwcJDq1qhRA7Vr18alS5cQEBCAv/76C99884102QbpDhM9yrH169dj5MiRaNu2LerUqYOWLVvi2bNncHR0xJdffomzZ88CAExNTdGtWzds375dWvf7779HeHi4DqOn7EpKSsLSpUvRtWtXTJ48GfHx8ejatSt69+6NLl26IDk5+b2J3oIFC+Dm5oZhw4bByckJv//+e5Y+WJJuGRsbo0ePHmq/tyNHjsSWLVvQv39/mJubo2nTplAoFGjdujUSEhLg6+sLAKhatSqKFSuGM2fOAACsra0xfPhw+Pv762x/6P3++usvdO7c+YN1vvvuO3h6eqJq1apwd3cHAAQGBmLYsGFYvHgxWrduDX9/f4wZMwZI/5B/5swZvH79GrVq1UKHDh0QExOTJ/tDWSeTyTBz5kx4e3tnWmf69OnYtGkTfvrpJ4wdOxZ9+vSBn5+fdA3e06dPAQAVK1aEq6srTp48CQCwtLTEyJEjERAQkGf7Q5qLiYmRfkfHjBmD/fv3S8v++8XN8ePHpS8HZDIZ9u/fj9DQULRo0QILFy5EjRo1sHHjRh3tCUkEUTb4+/uLJUuWiL///lsqa9CggTh48KD0+qeffhLfffedEEKIzZs3i44dO0rLvL29Rb169fI4atKmVq1aifbt24sdO3aIDh06iBYtWkjLevfuLQYMGCCuX78uGjVqpLZe9erVRa1atcSSJUtEaGioDiKn7FKpVOL+/fvizz//FFWrVhVCCJGamirMzMxEUFCQVM/d3V1s3LhRCCHEqFGjxIwZM6Rlv/zyixgxYoQOoqfsSExMFHZ2dtJxjY+PF69evZKWX7x4Ubi6ugq5XK623rRp00SXLl3Uym7cuCGCg4OFEELMnDlT/Pvvv3myD6S5+fPni8GDBwshhLh9+7ZYuHChuHXrlrS8fPnyYuvWraJdu3aicOHCwtPTUzx69EgIIcQPP/wgpk6dKtVdvny5GDp0qA72gjT177//Ck9PT2FtbS32798vhBBiz549olWrVlKdmJgYYWNjI6Kjo4UQQtSoUUOcP39eZzFT1jDRoyxZvXq1+OOPP0SpUqXEwIEDRYkSJcS+ffuEEEJ8/vnn4vDhw1LdY8eOicaNGwshhIiNjRU2NjYiPDxcWh4fH6+DPaD/unbtmvSh7fr16+/9MHb+/Hnxyy+/CH9/fyGEEM+fPxc2NjYiKSlJCCFEUlKScHFxEVevXhVCCKFQKETjxo1F586dRdOmTdW29fr16zzYK9KW6OhoUa9ePVG5cmXRsGFDIZPJxK1bt4RSqRSWlpbSB3khhJgxY4b4/vvvhUh/X5UvX15alpiYKFJTU3WyD5Q9gwcPFvPnzxfz588X5ubmolChQmLQoEEiNTVVbN68WXz11VcZ1nn9+rUoU6aMaNmypRg+fLioVKmSKFu2rDh79qxO9oE0ExQUJOzs7MS2bdtExYoVxcCBA4Wtra24f/++EEKIIkWKiMaNG4sNGzZkSPZv3bolypQpI71WKBQiJSUlz/eBNBMYGCiKFSsmlixZIqKioqTyxMREUaRIERESEiKVlS9fXqxbt04IIcTChQvF8OHDdRIzZR2nblKm3lxXAwDPnj1D//79cf78eaxfvx5bt27FjBkzAAAdO3bE0qVLkZKSAgD4+++/UbFiRSB92sb27dthYmIibcvc3DzP94UyWrhwIfbt2wek3w3zzbWTcrkcc+bMwbx58zBmzBhcunQJDRs2hFwuh5GREWQyGYyNjQEAJiYmaNq0qfRMRFNTUxw8eBAvXrxA8+bN1dqzsbHJ832krFMoFEhKSpJer1ixAiVLlsS///6LS5cuSbfINzAwQPv27bFgwQIAQHJyMs6ePSv9ztetWxezZs2SrtExMzPjg3PzMSEEnj17BqRPx/r9999x8uRJvHz5EqGhobh//z42btwINzc3/PPPP4iNjZXWlcvlsLGxwa1bt9C9e3dUrFgRW7ZswbNnz9CsWTMd7hVlRUpKCpKTkwEATk5OqF27NmbNmoXbt29j/fr1mDx5MhYuXAgAaN++PSpWrIj+/fujcOHCuHTpEkaPHo2EhATUqlUL8+bNk37nTU1NYWRkpNN9o/fz8/PD6tWrMW/ePAQHBwMA/P39UaxYMbi7u+P69es4ePAgIiMjYWZmhl69emH27NlITk7G2rVrYWNjI03jfzNlm/I5XWealLd+/PFHkZCQkOny+Ph4MXbsWFGkSBFhY2MjTce4d++esLa2FiqVSoj0KV2lS5cWd+/eFYmJicLd3V2UKlVKVKhQQdStW1ftGyDKn44cOSKaN28uvZ4/f75ISEgQSqVSODk5ibZt20rHu3///mLRokVCpE/B3LBhgxDpI7bly5cXJ0+e1NFeUE7ExcWJ0NBQMWnSJGFmZiYsLS2lY/ntt9+KX3/9Vap7584dUaxYMZGamir8/f1FuXLlRLVq1YSjo6Po3r27SE5O1uGekKZCQ0NFkSJFhEKhECqVSpQqVUqsWbNGWn7s2DHx+eefCyGE+Prrr0WtWrXE9OnTRffu3YW1tbV4/vy5DqOn7IqPjxehoaFi9OjRwsTERFhZWYnVq1cLkX6pxbszMcLDw4Wtra2Ij48XISEhok6dOqJo0aKiSJEiokqVKmL27NkZRvco/+rVq5ewt7cXgwcPFhMnThTly5cXiYmJQqVSieHDh4tatWqJFi1aCA8PD1GtWjUh0s8P1apVEwYGBqJdu3bi1atXvPSigGGi94lp3bq12L59e6bLhw4dKkaOHClCQkLEkSNHhJmZmfjnn3+EEELUrFlTnDt3Tqo7ZcoUMX78eCGEEEqlUty4cUPcu3dPSg4of0tJSRHFixcXAQEBQggh2rZtK3bs2CFE+hcC715XdeHCBVGjRg0hhBCXLl2SpvEUL15cjBo1Skd7QDm1YMECUbNmTTFo0CARGxsrjh49KsqWLSuUSqVYu3at6NChg1T35s2bAoA4duyYEOnTs/755x/h6+urwz2g7FKpVMLDw0O6zkak/114MxX/f//7n3SNtRBC3L17V1SoUEFad8uWLWLChAli8+bNIjIyUgd7QJo4f/68iIuLE0uWLBHVq1cXnp6eIiYmRty7d084ODiIe/fuvfdSiw4dOoht27YJkX78nz59yg/6BUBqaqq4c+eOePLkiVT24MEDoVAoxMGDB0XLli2FgYGBWLt2bYZ1b968KWxsbNQ+y31ogIDyNyZ6n5gdO3aIVq1aCZVKJTZs2CC6du0qfZsnhBClSpUSEyZMEM7OzqJatWpi/vz5IiwsTAghxKJFi8SgQYOkuk+fPhW9e/fWyX6Q5vz8/MSAAQOkb+3mzJkjhBBi586dok2bNkKkj+CWLFlSbQS3TJky4t69e0IIISIjI8WJEyeka/co/7t7965YsmSJWLp0qYiLixMi/bocAwMD8fTpU6mem5ubOH36tEhMTBTlypUTo0ePFuvWrRNVqlQR48aNUxvlo/wvJSVF7N69W7Rr1046zt27d5eusxFCiO3bt4tOnToJIYR4/PixMDMzEzt37hR3794VrVq1EiNHjtRZ/KS58PBwsXnzZuHl5SWqVq0qHj16JEJCQoShoaG4du2aVG/69Onihx9+EEII0adPH/Hbb79Jyw4ePChmz56tk/hJM1evXhXly5cXn332mahUqZKYMGGCtKxv376iatWqYvPmzWLt2rXSaH1qaqrYs2ePaN++vShWrJjaDfaoYGOi94lJSEgQNjY2wsvLS3Ts2FGsWbNGlC5dWrqZSrly5USvXr3E7du3M6wbHBws7OzshEKh0EHkpA0pKSmiYsWKYtKkSeLevXviu+++E5UrVxYi/b1hZ2cnfVtbs2ZNtTtqzZ8/X+zatUtnsVP2qVQqkZKSItzc3ETJkiXF999/L4YNGybc3d2lOu7u7mLv3r3S619//VV8/fXXQqT/zg8fPlz06NFDHD16VCf7QJoLCQkRVapUEc2bNxdbtmyRvpU/dOiQaNasmVQvJiZGmJiYiIiICCGEEPXq1RMeHh6icePGYtKkSbyBVgEVHR0trK2txVdffSWUSqVU3qpVK7Fp0ybp9datW0X37t2FEEL8/fffon79+jqJl3JOqVSKKlWqiH/++UfcuXNHjBo1ShgYGAhfX18RFRUlChUqJJ0H5s2bJwwNDYWvr69QKpWiV69eYuvWrfx91zNM9PScSqUSV69eFTdv3pTKvv76a1GlShVptGbfvn3iyy+/FCL97nkNGzYUz549E35+fmL58uWiUaNG0i23nz17pqM9IW14/PixcHFxkV6npqaKUqVKiRs3bgghhPjmm2/E0qVLhUi/o9ab221TwREbGytev34tunfvLo3E3L17VygUCrF27VpRr149AUBcvnxZCCHEpk2bpNEcIYR49eqVsLOz41QdPfDLL7+ozcJ4Izk5WRQrVkya1nXp0iVhZmYmVqxYIYQQ4vjx49I5gQqGO3fuiK5du4oaNWqI8ePHS3dG7tGjh5gyZYpa3e3bt4tSpUqJ27dvi/v374vatWuL5cuXC5H+N8HPz08n+0DZ4+vrKzp37izq1q0rfSnr6+srXF1dRfXq1YWzs7OYMGGC8PHxESL982CpUqXEpEmTxA8//CBq164tFi5cKK5cuaLjPaHcxERPTz169EhcuXJFfPHFF6Jy5crC0dFR9O3bVwghxOnTp6VRHJF+i/yiRYuKwMBAoVAoxMiRI4Wtra1wdHQUffr0EceOHePt0fVEZGSksLKyUrtxRqdOnaRb4585c0bUrl1biPTbpr+ZtksFR7NmzUSTJk3ErFmzpNF3hUIh6tevL7744gtx+PBhtWsw31yX82Y0Rwihdl0HFVw3btwQdnZ2ws3NTZQoUUIUK1ZMbNmyRQghxJIlS0S5cuVE//79pVkdBw4c0HXIpAF/f39hZ2cnfv/9d/HPP/+IZs2aidGjRwshhDh8+LD0uKM34uPjhaWlpfDw8BDVq1cXM2fO5N/4AuLq1ati2rRp4tq1a8Ld3V1s3rxZrFu3TpQoUULEx8eLpKQkYWpqKjZv3qw2ivvmi7t79+6JgQMHiilTpqg9J5P0FxM9PbVnzx5RpEgRMWnSJCHS765Xo0YNceDAAaFSqYSLi4v0fBwhhBgxYoSYN2+eDiOmvNK6dWvpbqr//vuvcHNzE8WKFRMpKSlCpVLx4cb5UEREhLh27ZqIjY39aN0VK1YIJycntbKLFy+KatWqSaP4/fr1E/b29lLC/9tvv/GmKnoqODhY3L59W4SEhIh9+/aJOnXqSMv27t0rVqxYoZbkU/72+vVrMX/+fLF27Vrp93nVqlWiW7duUp2goCBRqFAhkZiYKJKTk0XRokXFixcv1LYzYMAAXm9bgEyZMkWsWbNGVKtWTYwYMUJYWFiIadOmScvbtm0rfYkzbtw4UadOHXHw4EGxbt060a5dO+Hi4sJk/hPF5+gVcMnJyQgODkbXrl3h7OyMqVOnAgA6dOiA1NRUtGrVCgBgYWGBkSNH4o8//oBMJkOfPn2wdetWaTtDhw5F8eLFdbYflHdWr16NAwcOwNbWFl988QUWLVqEP//8E4aGhpDJZKhSpYquQ6R3jB8/HhUqVECvXr1Qrlw56ZlnmenZsyeio6PVnndWoUIFBAcHY9GiRRgwYAASExPxzTffIDQ0FAAwcuRIlCpVKtf3hfJeiRIlUK5cORw7dgwTJ05Er169pGXdunXDiBEjUKRIEZ3GSB8WGRmJBw8eQKFQoGnTpvD19cWvv/6K2bNnA+nPKJXL5VJ9Jycn2NnZwdfXF8bGxujZsye2bdumts1+/fphy5Yteb4vpJng4GBMnz4dV65cwYoVKzBu3DjpOXgA0L9/f+kz3fz589GtWzf89ttvOHfuHHr16oWHDx/yeaafKl1nmqS5hIQEUaJECdG2bVuxfv16cfPmTVG2bFlx5swZIYQQgwcPFjNmzJDqb9q0SXTt2lWI9JGcevXq6Sx20j1fX1/pOg7Kn1avXi26d+8uXRw/ZMgQMXbs2I+u16lTJ7Fx40a1shMnTogBAwaIRYsW8fq7T8iTJ0+EjY2N+Oqrr8Tp06d1HQ5lkUKhEGfOnBF79+4VDg4Ows7OTlSsWFF6DMaLFy+EnZ2dSExMFFFRUcLOzk7623/8+HFhb28vEhMThRBCBAQEZLhDslKpFKtXr+bjkAqIc+fOiTJlykiv/fz8hL29vfQ3PDExUdja2oqXL1/qMErKj4x0nWhS1j19+hSnT59GiRIl0LFjRxQqVAh169aFra0tBg4cCADw8vLCypUr0bx5c/Tr1w9t2rSBnZ0dzM3NMWXKFCxZsgQAUKVKFVy+fFnHe0S6xBGc/K9u3br4+uuvYWJiAgCoWLEigoKCPrpev379sHLlSnz99ddSWYsWLdCiRYtcjZfyn/Lly+PFixewtbXVdSiUDSkpKejSpQvq1KmDhw8fwsjICJUrV4aDgwMAoHTp0qhVqxaOHDmC7t27Y+PGjejUqRMKFy4MU1NT7N69G2ZmZgAAFxeXDNs3MDDA0KFD83y/KKOYmBhYW1t/sM7nn38OpVIJHx8fVK9eHSVLlkTVqlVx7NgxfPXVVzAzM8Pq1ashk8nyLG4qGDh1s4BYuHAh3N3d4ePjg5UrV+K7774D0j/QBQQESPW6d++OkydPIiYmBk2bNoWDgwPu37+PS5cuYd26dejRo4dU18CAh58oP5o3bx78/f1Rq1YtKckDIH2o+5j27dujVKlSUKlUuRwpFQRM8goeS0tLtGvXDp999hns7OxgZWWFgQMH4sCBA1Kdfv36SdP1OnbsiPDwcFy6dAnPnz/Hl19+qcPo6WOePHmCS5cuwdnZGfb29vjyyy8RGRmZaf3MLrkJDAyUXnfv3h0lSpTI9dipYOEn/XxIqVQiOjpaeh0YGIg9e/bg3LlzqFChAl69eoWDBw8iISEBHTp0wN27d/Hy5UsAgLW1NVq2bIm9e/dKJwZLS0usXbsWrVu31uFeEVFmEhMT1V5v374dCoVCrezGjRtQKpVo1KjRR7dnamqKdevW8cscogKsX79+uHjxotrrnTt3QqlUAgC6du0KpVIpfaFjamqKkiVLclQnnwoODoafnx8SExPRoEEDjB07FgcPHkR8fDzKly+PH3744YPr9+vXD7t375Ze9+7dG6NGjcqDyKkg46eAfEKpVOLJkyfYuXMn7OzsULx4cXh6ekKhUOD58+fw8/NDrVq1cOXKFcyaNQsBAQEwNzeHqakpunbtih07dkjbmjp1Kpo0aQKknxh27Ngh/WEgovwlPDwcFSpUwKVLl6QypVKZ4cL5mTNnwsvLCwDg6+uL8PDwPI+ViPKOh4cHAgIC8PTpUyD9pkolS5bE+fPngfRRv6NHj/ILnXzuyZMn6NSpE2rUqIELFy6gUKFC6NSpE0qWLIk6derAxMQECxcuxP79+zN8wfeuSpUq4caNG3kaOxV8PDvkE35+fmjatCmWL18OHx8fREZGIjo6GosWLUK1atUQFxeHq1evYufOnWjbti1UKhV8fHwAAN988w3i4uKkbVWtWhWVKlUC0q/p+euvv3i3JaJ84vbt2/Dy8oKXlxf8/PxQtGhRbN68Gd27d4evry/wnkTv3r17uHfvHl6+fImmTZuiQYMGuHXrVoZt+/r6YubMmZg4cWKe7hMRaZ+hoSF69eqlNl3v2LFjnJZZgMTExMDd3R2NGjXCy5cv0b9/fyD9S/gHDx5I9aysrFC4cGG8evXqg9srWrRorsdM+oWJng4kJiZi5cqVGDduHE6cOAEAKFu2LMqWLQsPDw+4urrCwsIC8+bNw+bNm2Fvb48xY8agffv2mD59Or755hs4OjpixowZAICGDRvi559/zrS9GjVq5Nm+EVHmDhw4gNatW8PCwgJyuRz169fHq1ev8OWXX2LGjBno0KEDYmJiMiR6hw8fRmhoKE6fPo2ffvoJL1++lB6dEh0djbVr1+Lzzz9HgwYNEB4ejm7duulwL4lIW4YPH47mzZtLr+3s7HQaD2XPhQsXULJkSfz0008wNTWVyps1a4bo6GjMnDkTQUFBmDNnDgoXLgxnZ+f3bic8PBy3b9/Ow8hJX8iEEELXQXxK4uLiULduXVSvXh3VqlXD77//jvnz52PAgAFYuXIlTp06hf379wMAXr9+DWdnZ8THxwMA/vjjD1y5cgVlypRBu3bt3nsnLSLKfQqFAuHh4XBxcUFcXBx8fHzQsGFDtToxMTE4c+YMKlWqhMqVKwMAGjdujDFjxkg3VBkyZAgcHByk52F5eXnh9u3bePjwIS5fvgwnJycAQEREBExMTGBlZZUhljc3WHpzl10jI95MmYgoPzh37hy+//573L17N8MyLy8vHDlyBObm5nB2dsb8+fNRoUIFablCocCRI0ewdetWXLx4Ed9//z2mT5+ex3tABZ6un+/wqdm4caNwd3eXXnt7ewtHR0ehUqlEZGSkMDMzE+vXrxfPnz8X33zzjWjfvr1O4yWijHbu3Cn9bvr7+4uhQ4dKyw4cOCD++ecfUbp0aeHu7i5sbW3F+fPnhRBC1KlTR5w6dUqqe/DgQdG8eXPptUqlEl27dhWVKlUS0dHRebpPRESkXUlJSaJ06dJi1apVIjU1VTx69EhMnTpVhIWFifv37wtXV9f3Pstw0qRJws7OTrRo0UJs2bJFxMXF6SR+Kvg4dTOPKRQK6Tk4SP+GPzk5Gb6+vrCzs0ObNm2wdu1a9OzZE0ZGRtiwYYNO4yWijDp16oTLly8jPDwcrq6uGDp0KG7evAkA+Ouvv9CpUyccOnQIp06dwu+//45p06YBAFq2bIl169ZJ27lz5w4cHR2l1zKZDLt378bDhw8/+lwlIiLK30xMTLB3714sXboUpqamaNmyJeRyOVQqFapWrYoiRYrg3LlzGdZr1aoV7t+/jxMnTqBfv36wsLDQSfxU8HHqZh57/PgxGjVqhIcPH8LBwQFBQUGoUKECQkNDYWVlhQMHDmD58uU4ffq0rkMlog8YNGgQPvvsM4waNQobN27E0aNHsW/fPnh7e6NLly7SRfWpqalwdnbGlStXYG1tjcaNG6NIkSIwMzPDs2fPcOrUKZQtW1bXu0NERLkoISEB5ubmamXnzp2Dk5MTypcvr7O4SL8x0dOBCRMmYPfu3WjWrBlOnjyJcePGYdy4cQCA5ORkODk54datW7wGjygf2r59O/766y8g/Yub69evIzY2Fq6urvD19YW1tTXKli2LAwcOoGbNmgCA77//Hg4ODpg8eTLi4uLw999/w8zMDO7u7jAzM9PxHhEREZE+YqKnI+fOncPz58/RtGlTtYtvAeDPP/9E/fr1eRtdonxm6dKlWLt2LWbNmgU/Pz9MnDgRd+/eRaVKldCrVy80b94cQ4cOxZQpU5CYmIhFixYB6Q87nzp1Ko4dO6brXSAiIqJPBBM9IqIsatiwIWbMmIEWLVoAAEaPHo3ChQtj9uzZOHr0KObNm4eLFy/iyZMnaNasGQIDA/kMSyIiItIJ3oyFiCiLLCwskJSUJL1u2rQptm3bBiEEWrVqhadPn8LX1xcVKlTA9u3bIZPJdBovERERfbqY6BERZVHXrl2xaNEiJCcnQ6VS4dixY1AoFLhw4QKMjIxw8OBB2NvbAwCaN28OAwOeYomIiEg3OHWTiCiLUlJS0L17d9y4cQMGBgZo0qQJRo8eDRcXF+nh5kRERET5ARM9IqJsunPnDkxMTFClShVdh0JERET0Xkz0iIiIiIiI9AwvICEiIiIiItIzTPSIiIiIiIj0DBM9IiIiIiIiPcNEj4iIiIiISM8w0SMiIiIiItIzTPSIiIiIiIj0DBM9IiIiIiIiPcNEj4iIiIiISM8w0SMiIiIiItIzTPSIiIiIiIj0DBM9IiIiIiIiPcNEj4iIiIiISM/8HwpUnfM2Go4gAAAAAElFTkSuQmCC", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
target recallrecall nullrecall p
alignmentX_harmonyX_pcaX_tvnX_harmonyX_pcaX_tvnX_harmonyX_pcaX_tvn
model
cp_measure0.0790.0890.1760.0550.0510.1270.0200.0100.010
dinov20.0550.0520.1040.0540.0490.0890.4460.3070.059
dinov2_random0.0940.1140.0870.1050.1090.1010.9210.2480.980
morphem0.0600.0690.1360.0630.0660.1090.6830.3560.010
openphenom0.0720.0570.1270.0810.0660.0910.8710.8510.010
subcell0.0570.0420.1540.0620.0570.1200.7030.9900.010
\n", "
" ], "text/plain": [ " target recall recall null recall p \\\n", "alignment X_harmony X_pca X_tvn X_harmony X_pca X_tvn X_harmony \n", "model \n", "cp_measure 0.079 0.089 0.176 0.055 0.051 0.127 0.020 \n", "dinov2 0.055 0.052 0.104 0.054 0.049 0.089 0.446 \n", "dinov2_random 0.094 0.114 0.087 0.105 0.109 0.101 0.921 \n", "morphem 0.060 0.069 0.136 0.063 0.066 0.109 0.683 \n", "openphenom 0.072 0.057 0.127 0.081 0.066 0.091 0.871 \n", "subcell 0.057 0.042 0.154 0.062 0.057 0.120 0.703 \n", "\n", " \n", "alignment X_pca X_tvn \n", "model \n", "cp_measure 0.010 0.010 \n", "dinov2 0.307 0.059 \n", "dinov2_random 0.248 0.980 \n", "morphem 0.356 0.010 \n", "openphenom 0.851 0.010 \n", "subcell 0.990 0.010 " ] }, "execution_count": 8, "metadata": {}, "output_type": "execute_result" } ], "source": [ "fig, ax = plt.subplots(figsize=(9, 4.5))\n", "order = list(blocks)\n", "width = 0.26\n", "for offset, alignment in zip((-width, 0, width), (\"X_pca\", \"X_tvn\", \"X_harmony\"), strict=True):\n", " block = comparison[comparison[\"alignment\"] == alignment].set_index(\"model\").loc[order]\n", " x = np.arange(len(order)) + offset\n", " ax.bar(x, block[\"target recall\"], width * 0.9, label=alignment)\n", " # the shuffled-annotation level for this exact representation\n", " ax.hlines(block[\"recall null\"], x - width * 0.45, x + width * 0.45, color=\"black\", lw=1.6)\n", "ax.plot([], [], color=\"black\", lw=1.6, label=\"shuffled annotation\")\n", "ax.axhline(0.10, color=\"crimson\", ls=\":\", lw=1, label=\"the 10% people assume\")\n", "ax.set_xticks(np.arange(len(order)), order, rotation=20, ha=\"right\", fontsize=8)\n", "ax.set_ylabel(\"recall of annotated pairs\")\n", "ax.legend(fontsize=7)\n", "fig.tight_layout()\n", "plt.show()\n", "\n", "comparison.pivot(index=\"model\", columns=\"alignment\", values=[\"target recall\", \"recall null\", \"recall p\"]).round(3)" ] }, { "cell_type": "markdown", "id": "c020", "metadata": {}, "source": [ "The black bars are chance for that representation, and they are not a constant 10%: for the trained embeddings and\n", "`cp_measure` they run from 5 to 8% without TVN and 9 to 13% with it, and for the untrained model they sit at\n", "10 to 11% under every alignment. Read every recall against the bar sitting on it, not\n", "against the red line.\n", "\n", "Once you do, the readout says something clean, and something different from what the raw numbers suggest.\n", "**`pp.tvn` lifts recall clearly above its own null on three of the four trained embeddings and on\n", "`cp_measure`, and on DINOv2 only marginally (p = 0.06, five of the 100 shuffles scoring as high).\n", "Under either of the other two alignments no embedding clears its own null, but `cp_measure` clears it under\n", "all three.** And the\n", "untrained model never clears it: its apparently high raw recall, the highest in the unaligned column, is\n", "entirely a property of its null, which is high because a near-random projection spreads pairs into the tails.\n", "That is exactly the trap a fixed baseline sets: in the unaligned and Harmony columns it would have made the\n", "untrained control look like the best model on the page.\n", "\n", "The shuffled annotation is what exposed the fixed baseline; the untrained model is where it would have done\n", "the most damage. Keeping one in the comparison is the cheapest check that a pipeline measures anything." ] }, { "cell_type": "markdown", "id": "c021", "metadata": {}, "source": [ "## The laboratories\n", "\n", "The same `metrics.pc_regression`, now on the source." ] }, { "cell_type": "code", "execution_count": 9, "id": "c022", "metadata": { "execution": { "iopub.execute_input": "2026-10-21T00:52:45.795485Z", "iopub.status.busy": "2026-10-21T00:52:45.795327Z", "iopub.status.idle": "2026-10-21T00:53:00.487862Z", "shell.execute_reply": "2026-10-21T00:53:00.486627Z" } }, "outputs": [ { "data": { "text/html": [ " \n", " \n", " " ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "
\n", "
" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
alignmentX_harmonyX_pcaX_tvn
model
cp_measure0.0050.0210.022
dinov20.0060.0300.045
dinov2_random0.0020.0190.010
morphem0.0080.0440.042
openphenom0.0040.0350.036
subcell0.0070.0340.027
\n", "
" ], "text/plain": [ "alignment X_harmony X_pca X_tvn\n", "model \n", "cp_measure 0.005 0.021 0.022\n", "dinov2 0.006 0.030 0.045\n", "dinov2_random 0.002 0.019 0.010\n", "morphem 0.008 0.044 0.042\n", "openphenom 0.004 0.035 0.036\n", "subcell 0.007 0.034 0.027" ] }, "execution_count": 9, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# pl.metrics groups by \"metric\" and draws one bar per \"representation\", which here is one group per\n", "# feature set and one bar per alignment.\n", "mt.pl.metrics(\n", " comparison.rename(columns={\"model\": \"metric\", \"alignment\": \"representation\", \"source variance\": \"value\"}).assign(\n", " better=\"lower\"\n", " )[[\"metric\", \"representation\", \"value\", \"better\"]]\n", ")\n", "comparison.pivot(index=\"model\", columns=\"alignment\", values=\"source variance\").round(3)" ] }, { "cell_type": "markdown", "id": "c023", "metadata": {}, "source": [ "Harmony takes the laboratory down to a fraction of a percent on every feature set. TVN does not\n", "consistently reduce it: it is unchanged on three blocks, up by half on DINOv2, and down on SubCell and the\n", "untrained model. That is consistent with what the two do (Harmony mixes the batches in every local\n", "neighbourhood, while TVN aligns each batch's control covariance to the pooled one) and a plausible reason the\n", "two readouts disagree.\n", "\n", "Read this together with the tables above, not instead of them. TVN leaves the laboratory in place and is still\n", "the only alignment under which the embeddings' annotation recall clears its own null; Harmony removes it and is the only one under which most blocks\n", "call many compounds. A correction that removes the batch may also have taken biology with it, and one that leaves it\n", "may still help, so neither readout alone tells you which happened. Which alignment to use is a decision about\n", "what the screen is for, not a default." ] }, { "cell_type": "markdown", "id": "c024", "metadata": {}, "source": [ "## What the leading components are\n", "\n", "`pc_regression`, per component, on the cell count and on the laboratory." ] }, { "cell_type": "code", "execution_count": 10, "id": "c025", "metadata": { "execution": { "iopub.execute_input": "2026-10-21T00:53:00.491109Z", "iopub.status.busy": "2026-10-21T00:53:00.490942Z", "iopub.status.idle": "2026-10-21T00:53:01.625155Z", "shell.execute_reply": "2026-10-21T00:53:01.624013Z" } }, "outputs": [ { "data": { "text/html": [ " \n", " \n", " " ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "
\n", "
" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ " \n", " \n", " " ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "
\n", "
" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ " \n", " \n", " " ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "
\n", "
" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ " \n", " \n", " " ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "
\n", "
" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ " \n", " \n", " " ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "
\n", "
" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ " \n", " \n", " " ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "
\n", "
" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
keycell countsourcefirst component, r with count
openphenom0.3230.035-0.696
dinov20.3760.0300.744
dinov2_random0.0150.019-0.116
subcell0.2160.0340.702
morphem0.3010.0440.736
cp_measure0.0600.021-0.176
\n", "
" ], "text/plain": [ "key cell count source first component, r with count\n", "openphenom 0.323 0.035 -0.696\n", "dinov2 0.376 0.030 0.744\n", "dinov2_random 0.015 0.019 -0.116\n", "subcell 0.216 0.034 0.702\n", "morphem 0.301 0.044 0.736\n", "cp_measure 0.060 0.021 -0.176" ] }, "execution_count": 10, "metadata": {}, "output_type": "execute_result" } ], "source": [ "fig, axes = plt.subplots(2, 3, figsize=(14, 6.5), sharey=True)\n", "for ax, (model, wells) in zip(axes.ravel(), blocks.items(), strict=True):\n", " mt.pl.batch_variance(wells, keys=[\"Metadata_CellCount\", \"Metadata_Source\"], use_rep=\"X_pca\", n_comps=10, ax=ax)\n", " ax.set_title(model, fontsize=10)\n", "fig.tight_layout()\n", "plt.show()\n", "\n", "variance = pd.concat(\n", " {\n", " model: mt.metrics.batch_variance_explained(\n", " wells, keys=[\"Metadata_CellCount\", \"Metadata_Source\"], use_rep=\"X_pca\"\n", " ).set_index(\"key\")[\"value\"]\n", " for model, wells in blocks.items()\n", " },\n", " axis=1,\n", ").T.rename(columns={\"Metadata_CellCount\": \"cell count\", \"Metadata_Source\": \"source\"})\n", "variance[\"first component, r with count\"] = {\n", " model: float(np.corrcoef(wells.obsm[\"X_pca\"][:, 0], wells.obs[\"Metadata_CellCount\"])[0, 1])\n", " for model, wells in blocks.items()\n", "}\n", "variance.round(3)" ] }, { "cell_type": "markdown", "id": "c026", "metadata": {}, "source": [ "On every trained embedding the first component correlates with the cell count at about 0.70 to 0.74 in\n", "magnitude, against 0.18 for `cp_measure` and 0.12 for the untrained model, and over the fifty components the\n", "count accounts for between six and twelve and a half times as much of the variance as the laboratory does. The\n", "largest axis of these models tracks how full the well is. The untrained model does not do it, so the trained\n", "models learned it; `cp_measure`, a per-cell measurement summarised over the well, does it far less.\n", "\n", "The untreated control wells differ 1.7-fold in cell count between sites, as the\n", "[dataset page](../datasets/jump_lite.ipynb) shows. If a good part of what separates the laboratories is how\n", "dense their wells are, a representation whose largest axis tracks density entangles the batch with the biology\n", "on that axis, which would be one reason the embeddings need the correction more than `cp_measure` does. This\n", "page does not test that." ] }, { "cell_type": "markdown", "id": "c244f27a", "metadata": {}, "source": [ "## How much of CellProfiler each embedding carries\n", "\n", "A learned embedding has no `var` to read a hit off: a compound that stands out is an ID, not \"high texture in\n", "the mitochondrial channel\". So the question the comparison has skirted is what a black-box embedding actually\n", "holds of the named measurements it is being scored against. `metrics.variance_carried` answers it directly. It\n", "cross-fits a ridge from the embedding to each `cp_measure` feature and reports the out-of-fold R^2, averaged\n", "over each feature family: near one means the embedding linearly reconstructs that family, near zero means it\n", "does not carry it. The two blocks are the same 1,536 wells, so the metric aligns them on `obs_names` and needs\n", "nothing else. `classical` is the block to recover, normalized but not selected, so it still names every\n", "feature and its `var[\"feature_group\"]`." ] }, { "cell_type": "code", "execution_count": 11, "id": "ffcbab7b", "metadata": { "execution": { "iopub.execute_input": "2026-10-21T00:53:01.628129Z", "iopub.status.busy": "2026-10-21T00:53:01.627976Z", "iopub.status.idle": "2026-10-21T00:56:29.415313Z", "shell.execute_reply": "2026-10-21T00:56:29.414101Z" } }, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
feature_groupferrettextureradial_distributionintensityzernikeradial_zernikessizeshape
openphenom0.6100.5660.4960.4490.3750.2700.256
dinov20.6830.5000.4900.4290.4480.3020.312
dinov2_random0.1320.2200.1640.1180.1630.1010.013
subcell0.6840.5790.5160.4690.4500.3070.303
morphem0.7360.5710.5220.4980.4570.3260.331
\n", "
" ], "text/plain": [ "feature_group ferret texture radial_distribution intensity zernike \\\n", "openphenom 0.610 0.566 0.496 0.449 0.375 \n", "dinov2 0.683 0.500 0.490 0.429 0.448 \n", "dinov2_random 0.132 0.220 0.164 0.118 0.163 \n", "subcell 0.684 0.579 0.516 0.469 0.450 \n", "morphem 0.736 0.571 0.522 0.498 0.457 \n", "\n", "feature_group radial_zernikes sizeshape \n", "openphenom 0.270 0.256 \n", "dinov2 0.302 0.312 \n", "dinov2_random 0.101 0.013 \n", "subcell 0.307 0.303 \n", "morphem 0.326 0.331 " ] }, "execution_count": 11, "metadata": {}, "output_type": "execute_result" } ], "source": [ "carried = pd.DataFrame(\n", " {\n", " # blocks[model] holds the embedding in obsm[\"X_pca\"]; classical.X holds the named cp_measure block.\n", " model: mt.metrics.variance_carried(\n", " blocks[model], classical, use_rep=\"X_pca\", groupby=\"feature_group\"\n", " ).set_index(\"feature_group\")[\"variance_carried\"]\n", " for model in blocks\n", " if model != \"cp_measure\" # recovering cp_measure from itself is not the question\n", " }\n", ").T\n", "carried = carried[carried.mean().sort_values(ascending=False).index] # feature families, best-carried first\n", "\n", "fig, ax = plt.subplots(figsize=(9, 3.5))\n", "im = ax.imshow(carried.to_numpy(), cmap=\"viridis\", vmin=0, vmax=1, aspect=\"auto\")\n", "ax.set_xticks(np.arange(carried.shape[1]), carried.columns, rotation=30, ha=\"right\", fontsize=8)\n", "ax.set_yticks(np.arange(carried.shape[0]), carried.index, fontsize=8)\n", "for row in range(carried.shape[0]):\n", " for col in range(carried.shape[1]):\n", " value = carried.iat[row, col]\n", " # white on the dark low end, black on the bright high end, so every cell stays readable\n", " ax.text(\n", " col, row, f\"{value:.2f}\", ha=\"center\", va=\"center\", fontsize=7, color=\"white\" if value < 0.5 else \"black\"\n", " )\n", "fig.colorbar(im, ax=ax, label=\"out-of-fold $R^2$\")\n", "fig.tight_layout()\n", "plt.show()\n", "\n", "carried.round(3)" ] }, { "cell_type": "markdown", "id": "254f5aed", "metadata": {}, "source": [ "The trained embeddings carry the ferret family of `cp_measure` best (out-of-fold R^2 0.61 to 0.74) and texture next (0.50 to 0.58), an image model\n", "reconstructing how large each object is and how patterned each channel is, and the zernike and shape families\n", "less (0.26 to 0.46), since those turn on the exact segmentation the embedding never performed. `dinov2_random` carries little\n", "of any family: the untrained control sits well below the trained models everywhere, which is the point. A\n", "learned embedding stops being a black box once you can say what of the classical block it linearly rebuilds;\n", "read each row against the control, not on its own." ] }, { "cell_type": "markdown", "id": "c027", "metadata": {}, "source": [ "## Is the hit a hit, or did the cells die?\n", "\n", "This is the question a screener asks first, and on an embedding it is sharper than usual. Call hits, then ask\n", "`tl.cytotoxicity` whether the compounds furthest from the controls are the ones that lost cells." ] }, { "cell_type": "code", "execution_count": 12, "id": "c028", "metadata": { "execution": { "iopub.execute_input": "2026-10-21T00:56:29.419500Z", "iopub.status.busy": "2026-10-21T00:56:29.419328Z", "iopub.status.idle": "2026-10-21T00:56:30.891784Z", "shell.execute_reply": "2026-10-21T00:56:30.890741Z" } }, "outputs": [ { "data": { "text/html": [ " \n", " \n", " " ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "
\n", "
" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
rho(distance, viability)suspectgroups
model
openphenom-0.65564302
dinov2-0.69164302
dinov2_random-0.56864302
subcell-0.72064302
morphem-0.70064302
cp_measure-0.68864302
\n", "
" ], "text/plain": [ " rho(distance, viability) suspect groups\n", "model \n", "openphenom -0.655 64 302\n", "dinov2 -0.691 64 302\n", "dinov2_random -0.568 64 302\n", "subcell -0.720 64 302\n", "morphem -0.700 64 302\n", "cp_measure -0.688 64 302" ] }, "execution_count": 12, "metadata": {}, "output_type": "execute_result" } ], "source": [ "toxicity = []\n", "for model, wells in blocks.items():\n", " scratch = wells.copy()\n", " mt.tl.hit_calling(scratch, use_rep=\"X_pca\", n_permutations=200, seed=0)\n", " mt.tl.cytotoxicity(scratch, site_key=None) # one field per well, so there is no field count to divide by\n", " table = scratch.uns[\"mantispy\"][\"cytotoxicity\"]\n", " toxicity.append(\n", " {\n", " \"model\": model,\n", " \"rho(distance, viability)\": float(table[[\"distance\", \"viability\"]].corr(method=\"spearman\").iloc[0, 1]),\n", " \"suspect\": int(table[\"suspect\"].sum()),\n", " \"groups\": len(table),\n", " }\n", " )\n", " if model == \"openphenom\":\n", " mt.pl.cytotoxicity(scratch)\n", "pd.DataFrame(toxicity).set_index(\"model\").round(3)" ] }, { "cell_type": "markdown", "id": "c029", "metadata": {}, "source": [ "The `suspect` count is identical on every block, the untrained one included, so here it follows viability\n", "alone. The rank correlation between distance and viability is negative on every block, from −0.57 for the\n", "untrained model to −0.72 for SubCell, and `cp_measure` sits at −0.69, inside the range of the trained\n", "embeddings. Cell loss moves a profile away from the controls whichever feature block measures it, because a\n", "compound that kills cells has changed the well. The top of this screen's hit list is partly a cytotoxicity\n", "ranking whichever block you use." ] }, { "cell_type": "markdown", "id": "c030", "metadata": {}, "source": [ "### Should the cell count be removed?\n", "\n", "The reflex is to take the count out: drop the wells under a floor, or regress the count out of the profiles.\n", "Both are measured here on OpenPhenom, against keeping every well." ] }, { "cell_type": "code", "execution_count": 13, "id": "c031", "metadata": { "execution": { "iopub.execute_input": "2026-10-21T00:56:30.894534Z", "iopub.status.busy": "2026-10-21T00:56:30.894333Z", "iopub.status.idle": "2026-10-21T00:56:50.477422Z", "shell.execute_reply": "2026-10-21T00:56:50.476302Z" } }, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
wellscell count variancemAPsignificanttarget recallrecall nullrecall p
treatmentalignment
noneX_pca15360.3230.01600.0570.0660.851
X_tvn15360.3230.01830.1270.0910.010
X_harmony15360.3230.03970.0720.0810.871
floorX_pca13970.2500.01600.0480.0731.000
X_tvn13970.2500.02030.1520.1110.010
X_harmony13970.2500.034120.0530.0650.931
regressX_pca15360.2240.01500.0720.0680.307
X_tvn15360.2240.01620.1440.0990.010
X_harmony15360.2240.03230.0770.0750.406
\n", "
" ], "text/plain": [ " wells cell count variance mAP significant \\\n", "treatment alignment \n", "none X_pca 1536 0.323 0.016 0 \n", " X_tvn 1536 0.323 0.018 3 \n", " X_harmony 1536 0.323 0.039 7 \n", "floor X_pca 1397 0.250 0.016 0 \n", " X_tvn 1397 0.250 0.020 3 \n", " X_harmony 1397 0.250 0.034 12 \n", "regress X_pca 1536 0.224 0.015 0 \n", " X_tvn 1536 0.224 0.016 2 \n", " X_harmony 1536 0.224 0.032 3 \n", "\n", " target recall recall null recall p \n", "treatment alignment \n", "none X_pca 0.057 0.066 0.851 \n", " X_tvn 0.127 0.091 0.010 \n", " X_harmony 0.072 0.081 0.871 \n", "floor X_pca 0.048 0.073 1.000 \n", " X_tvn 0.152 0.111 0.010 \n", " X_harmony 0.053 0.065 0.931 \n", "regress X_pca 0.072 0.068 0.307 \n", " X_tvn 0.144 0.099 0.010 \n", " X_harmony 0.077 0.075 0.406 " ] }, "execution_count": 13, "metadata": {}, "output_type": "execute_result" } ], "source": [ "def remedy(treatment):\n", " \"\"\"The recipe `prepared` ran, with a cell-count floor or with the count regressed out.\"\"\"\n", " if treatment == \"none\":\n", " return blocks[\"openphenom\"] # already built above; rebuilding it is how the two quietly diverge\n", " wells = mt.ds.jump_lite(\"openphenom\")\n", " if treatment == \"floor\":\n", " wells = wells[wells.obs[\"Metadata_CellCount\"].to_numpy(dtype=float) >= 50].copy()\n", " mt.pp.normalize(wells, method=\"mad_robustize\", by=\"Metadata_Plate\", reference=\"negcon\")\n", " # The same step `prepared` runs. It drops nothing from this embedding, and leaving it out is how the\n", " # floor and regress arms would end up compared against a \"none\" arm built a different way.\n", " mt.pp.feature_select(wells, operations=(\"drop_degenerate\", \"drop_na_columns\"), na_cutoff=0.0)\n", " wells = mt.pp.subset_features(wells)\n", " if treatment == \"regress\":\n", " mt.pp.regress_out(wells, keys=(\"Metadata_CellCount\",), by=\"Metadata_Plate\")\n", " return align(wells)\n", "\n", "\n", "remedies = []\n", "for treatment in (\"none\", \"floor\", \"regress\"):\n", " trial = remedy(treatment)\n", " left = float(mt.metrics.pc_regression(trial, key=\"Metadata_CellCount\", use_rep=\"X_pca\")[\"value\"].iloc[0])\n", " for rep in (\"X_pca\", \"X_tvn\", \"X_harmony\"):\n", " result = scored(treatment, trial, rep)\n", " remedies.append(\n", " {\n", " \"treatment\": treatment,\n", " \"alignment\": rep,\n", " \"wells\": trial.n_obs,\n", " \"cell count variance\": left,\n", " \"mAP\": result[\"mAP\"],\n", " \"significant\": result[\"significant\"],\n", " \"target recall\": result[\"target recall\"],\n", " \"recall null\": result[\"recall null\"],\n", " \"recall p\": result[\"recall p\"],\n", " }\n", " )\n", "pd.DataFrame(remedies).set_index([\"treatment\", \"alignment\"]).round(3)" ] }, { "cell_type": "markdown", "id": "c032", "metadata": {}, "source": [ "Each remedy buys something the other does not. Regressing the count out takes its share of the variance from\n", "0.32 to 0.22 and puts annotation recall nominally above its own null under every alignment, but only under TVN\n", "is the difference clear (p = 0.01), as it is when keeping every well. It costs retrieval: under Harmony the compounds called drop from seven to three. The\n", "floor discards 139 wells, raises the calls under Harmony from seven to twelve at a small cost in mAP, and leaves\n", "recall below its null there. Under TVN neither moves the calls by more than one.\n", "\n", "In a compound screen the count is partly a treatment effect, since the compound reduced it, so regressing it\n", "out removes part of the phenotype and not only a nuisance. The JUMP recipe regresses the cell count out of the\n", "ORF and CRISPR arms and not out of compounds {cite:p}`Chandrasekaran_2023`. This page keeps every well and does\n", "not regress; report the count beside the readouts instead." ] }, { "cell_type": "markdown", "id": "c033", "metadata": {}, "source": [ "## Summary\n", "\n", "- On telling a compound from DMSO the blocks were close; on telling compounds apart across laboratories\n", " without a correction the embeddings called at most five compounds to `cp_measure`'s 32. Harmony narrowed that\n", " for every trained embedding but OpenPhenom.\n", "- Read Harmony as a range over seeds, and read `cp_measure`'s row as one recipe: the rank transform cuts its\n", " calls under Harmony from 95 to 55.\n", "- Measure chance with `known_relationships(n_permutations=...)`, and keep an untrained model in the comparison.\n", " Against a fixed 10% it would have looked like the best model without TVN.\n", "- The trained embeddings' largest axis tracks confluence, and distance from the controls tracks cell loss on\n", " every block.\n", "- The three alignments disagree and none dominates: Harmony lets the embeddings' compounds reproduce across\n", " sites, TVN lets their annotation recall clear its null.\n", "\n", "All of it is for off-the-shelf embeddings on lossy JUMP images and one CellProfiler recipe." ] }, { "cell_type": "markdown", "id": "n5452e74", "metadata": {}, "source": [ "## Next steps\n", "\n", "`tl.map` scores how well each perturbation retrieves its own replicates, and says nothing about which perturbations are alike or how far they sit from the controls.\n", "From here:\n", "\n", "- {func}`~mantispy.tl.similarity` stores the pairwise similarity of profiles, so you can ask which perturbations look alike.\n", "- {func}`~mantispy.tl.edistance` measures the distance of each group from the controls, or between every pair of groups, on the whole distribution.\n", "- {func}`~mantispy.tl.effect_size` reports which features moved, and by how much.\n", "- {func}`~mantispy.tl.consensus` collapses replicates into one profile per perturbation, which `tl.map(mode=\"consistency\")` and gene-level analysis read.\n", "\n", "See {doc}`../tutorials/profiles/choosing_a_readout` for which readout to report when." ] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.13.15" } }, "nbformat": 4, "nbformat_minor": 5 }