{ "cells": [ { "cell_type": "markdown", "id": "2826b5c2", "metadata": { "language": "markdown" }, "source": [ "# Ranking receptors for a fixed epitope \u2014 the 22-cohort VDJdb panel\n", "\n", "Which of these T-cell receptors reads this epitope? The input is a folder of co-folded complexes and\n", "nothing else: no binding assay, no alignment to a known binder, and no number the generator emits.\n", "\n", "This notebook runs the whole path \u2014 fetch, featurise, score \u2014 over the **balanced VDJdb receptor\n", "benchmark**: 1,089 TCRmodel2 complexes over 22 epitope cohorts, 523 real VDJdb binders against 566\n", "mock mispairings. It then reports the ranking **one cohort at a time**, because a macro average over\n", "22 cohorts hides both of the facts that matter: a macro number cannot tell two strong cohorts and one\n", "inverted cohort apart from twenty-two mediocre ones.\n", "\n", "The covariate the result turns on is **template coverage** \u2014 whether *some* receptor, not necessarily\n", "one appearing here, has already been co-crystallised with that peptide, and so whether the generator\n", "had a template for the interface it was asked to build. Six of the 22 cohorts are covered and 16 are\n", "free, and the two strata are never pooled.\n", "\n", "**Needs**: the MHC allele reference (`tcren build-mhc-ref`, once) and about 4 GB of disk for the\n", "structures. The featurisation pass is the expensive step; everything after it is arithmetic.\n" ] }, { "cell_type": "code", "execution_count": 1, "id": "27fa81ef", "metadata": { "execution": { "iopub.execute_input": "2026-09-02T02:57:42.252028Z", "iopub.status.busy": "2026-09-02T02:57:42.251718Z", "iopub.status.idle": "2026-09-02T02:57:43.096523Z", "shell.execute_reply": "2026-09-02T02:57:43.096215Z" }, "language": "python" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "python 3.12.13\n", "tcren 3.0.0\n", "numpy 2.5.1\n", "polars 1.43.0\n", "scikit 1.9.0\n" ] } ], "source": [ "# Environment. Nothing here is stochastic -- there is no seed to set, because no score in this\n", "# notebook samples: the transform, the class means and the covariance are all frozen in the wheel.\n", "import platform, sys\n", "\n", "import numpy as np\n", "import polars as pl\n", "import sklearn\n", "\n", "import tcren\n", "\n", "print(f\"python {platform.python_version()}\")\n", "print(f\"tcren {tcren.__version__}\")\n", "print(f\"numpy {np.__version__}\")\n", "print(f\"polars {pl.__version__}\")\n", "print(f\"scikit {sklearn.__version__}\")\n" ] }, { "cell_type": "markdown", "id": "8410222b", "metadata": { "language": "markdown" }, "source": [ "## 1 \u00b7 Bootstrap the deposit\n", "\n", "The benchmark ships as one archive plus a `metadata.tsv` on the Hugging Face dataset\n", "[`isalgo/tcren_structures`](https://huggingface.co/datasets/isalgo/tcren_structures), under\n", "`vdjdb_binder_benchmark/`. The metadata carries the binder label `y`, the epitope and allele, the\n", "generator's own `iptm` and `plddt`, and `native_structure_exists` \u2014 the template flag.\n", "\n", "The cell is idempotent: it downloads and unpacks only what is missing." ] }, { "cell_type": "code", "execution_count": 2, "id": "a7b2c8fa", "metadata": { "execution": { "iopub.execute_input": "2026-09-02T02:57:43.097869Z", "iopub.status.busy": "2026-09-02T02:57:43.097760Z", "iopub.status.idle": "2026-09-02T02:57:43.107998Z", "shell.execute_reply": "2026-09-02T02:57:43.107617Z" }, "language": "python" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "structures on disk 1089\n", "metadata rows 1089\n", "binders / mock negatives 523 / 566\n", "epitope cohorts 22\n", "template-covered rows 297 of 1089\n" ] } ], "source": [ "# Fetch and unpack the 1,089-structure deposit into ../data (gitignored).\n", "import os, tarfile, time\n", "from pathlib import Path\n", "\n", "from tcren.paper.bootstrap import fetch_hf_structures\n", "\n", "DATA = Path(os.environ.get(\"TCREN_NB_DATA\", \"data\"))\n", "SET = DATA / \"vdjdb_binder_benchmark\"\n", "PDBS = SET / \"positives\"\n", "\n", "if not PDBS.is_dir():\n", " fetch_hf_structures(DATA, folders=(\"vdjdb_binder_benchmark\",))\n", " with tarfile.open(SET / \"vdjdb_binder_benchmark.tar.gz\") as tar:\n", " tar.extractall(SET)\n", "\n", "meta = pl.read_csv(SET / \"metadata.tsv\", separator=\"\\t\", infer_schema_length=None)\n", "n_pdb = len(list(SET.glob(\"*/*.pdb\")))\n", "print(f\"structures on disk {n_pdb}\")\n", "print(f\"metadata rows {meta.height}\")\n", "print(f\"binders / mock negatives {int(meta['y'].sum())} / {int((1 - meta['y']).sum())}\")\n", "print(f\"epitope cohorts {meta['epitope'].n_unique()}\")\n", "print(f\"template-covered rows {int(meta['native_structure_exists'].sum())} \"\n", " f\"of {meta.height}\")\n" ] }, { "cell_type": "markdown", "id": "44081e96", "metadata": { "language": "markdown" }, "source": [ "## 2 \u00b7 Featurise, once\n", "\n", "`tcren features` is the expensive half of the pipeline: parse, annotate the TCR and MHC chains, build\n", "the contact map, and compute all 164 catalogued descriptors. It is run once and its table is re-scored\n", "for free.\n", "\n", "The exact command, which is what regenerates the cached table:\n", "\n", "```bash\n", "tcren features -s \"data/vdjdb_binder_benchmark/*/*.pdb\" \\\n", " -i placement,interface,topology,energetics,potts,kinetics \\\n", " -t 0 -o data/vdjdb_binder_benchmark/features.tsv\n", "```\n", "\n", "`-t 0` uses every core. All six families are asked for, because the five score channels are marginals\n", "of the descriptor families and `mechanics` needs `kinetics`. Measured here: **1,089 structures in 169 s\n", "on 16 cores** (1,967 s of CPU); on four cores expect roughly twelve minutes. The cell below skips the pass\n", "entirely when the table is already on disk, so re-running the notebook is cheap.\n", "\n", "Every table `tcren features` writes carries a `.provenance.json` beside it recording the version, the\n", "invocation and a SHA-256 digest of the descriptor catalogue. `tcren assess` refuses a table written\n", "under a different catalogue rather than mixing generations." ] }, { "cell_type": "code", "execution_count": 3, "id": "3dc90c8c", "metadata": { "execution": { "iopub.execute_input": "2026-09-02T02:57:43.109015Z", "iopub.status.busy": "2026-09-02T02:57:43.108943Z", "iopub.status.idle": "2026-09-02T03:00:32.614976Z", "shell.execute_reply": "2026-09-02T03:00:32.614383Z" }, "language": "python" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "featurised 1089 structures in 169 s\n", "1089 rows x 164 descriptors\n", "metadata joined onto 1089 of 1089 rows\n" ] } ], "source": [ "# Descriptors, one row per structure. Skipped when the cached table is already present.\n", "import subprocess\n", "\n", "FEATURES = SET / \"features.tsv\"\n", "FAMILIES = \"placement,interface,topology,energetics,potts,kinetics\"\n", "\n", "if not FEATURES.exists():\n", " t0 = time.time()\n", " subprocess.run(\n", " [sys.executable, \"-m\", \"tcren\", \"features\", \"-s\", f\"{SET}/*/*.pdb\",\n", " \"-i\", FAMILIES, \"-t\", \"0\", \"-o\", str(FEATURES)],\n", " check=True, capture_output=True, text=True)\n", " print(f\"featurised {n_pdb} structures in {time.time() - t0:.0f} s\")\n", "\n", "feats = pl.read_csv(FEATURES, separator=\"\\t\", infer_schema_length=None)\n", "descriptor_cols = [c for c in feats.columns if c not in meta.columns and c != \"complex.id\"]\n", "print(f\"{feats.height} rows x {len(descriptor_cols)} descriptors\")\n", "print(f\"metadata joined onto {int(feats['y'].is_not_null().sum())} of {feats.height} rows\")\n" ] }, { "cell_type": "markdown", "id": "d14ae964", "metadata": { "language": "markdown" }, "source": [ "## 3 \u00b7 Score\n", "\n", "`tcren.score.score_table` returns every read-out of the frozen model for this table. All of them are\n", "defined for **one** structure: the transform, the class means and the covariance were fitted on a\n", "hold-out that ships inside the wheel, so nothing is estimated from the rows being scored and a score\n", "does not move depending on what was scored beside it.\n", "\n", "| read-out | tier | what is estimated |\n", "|---|---|---|\n", "| `peptide_score` | 0 | nothing; the direction is fixed by the potential |\n", "| `pose_score` | 1 | a covariance over hold-out binders \u2014 no negative, no label |\n", "| `confidence_residual` | 1 | the same covariance, read as a conditional mean |\n", "| `binder_score` | 2 | class means and covariances, from hold-out binder labels |\n", "| `channel_*` | 2 | the same object, marginalized to one descriptor family |\n", "\n", "`binder_iptm` is `binder_score + logit(ipTM)`: two log-odds added, no coefficient to fit, still defined\n", "for a single structure. Higher is better throughout.\n", "\n", "From the command line the same table comes out of `tcren assess --features features.tsv -o scores.tsv`." ] }, { "cell_type": "code", "execution_count": 4, "id": "01c79b99", "metadata": { "execution": { "iopub.execute_input": "2026-09-02T03:00:32.616526Z", "iopub.status.busy": "2026-09-02T03:00:32.616423Z", "iopub.status.idle": "2026-09-02T03:00:32.669578Z", "shell.execute_reply": "2026-09-02T03:00:32.669136Z" }, "language": "python" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "model: tcren 3.0.0, 6429 binders / 1155 non-binders over 31 hold-out epitopes\n", "scored columns: ['pose_score', 'binder_score', 'channel_placement', 'channel_interface', 'channel_shape', 'channel_energetics', 'channel_mechanics', 'peptide_score', 'confidence_residual', 'binder_iptm']\n" ] }, { "data": { "text/html": [ "
\n", "shape: (5, 8)
epitopeycoverediptmpose_scorebinder_scorebinder_iptmconfidence_residual
stri64boolf64f64f64f64f64
"QAKWRLQTL"0false0.8393508-337.251593-37.842933-36.189528-0.234271
"RPIIRPATL"0false0.8676005-286.853361-23.749557-21.86965-0.358769
"FLRGRAYGL"0true0.631656-649.034789-115.492025-114.952699-1.157456
"QAKWRLQTL"0false0.7751346-355.210578-32.535676-31.298142-0.095934
"TPRVTGGGAM"0false0.8708996-390.091896-46.326726-44.417790.877601
" ], "text/plain": [ "shape: (5, 8)\n", "\u250c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2510\n", "\u2502 epitope \u2506 y \u2506 covered \u2506 iptm \u2506 pose_score \u2506 binder_score \u2506 binder_iptm \u2506 confidence_r \u2502\n", "\u2502 --- \u2506 --- \u2506 --- \u2506 --- \u2506 --- \u2506 --- \u2506 --- \u2506 esidual \u2502\n", "\u2502 str \u2506 i64 \u2506 bool \u2506 f64 \u2506 f64 \u2506 f64 \u2506 f64 \u2506 --- \u2502\n", "\u2502 \u2506 \u2506 \u2506 \u2506 \u2506 \u2506 \u2506 f64 \u2502\n", "\u255e\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2561\n", "\u2502 QAKWRLQTL \u2506 0 \u2506 false \u2506 0.8393508 \u2506 -337.251593 \u2506 -37.842933 \u2506 -36.189528 \u2506 -0.234271 \u2502\n", "\u2502 RPIIRPATL \u2506 0 \u2506 false \u2506 0.8676005 \u2506 -286.853361 \u2506 -23.749557 \u2506 -21.86965 \u2506 -0.358769 \u2502\n", "\u2502 FLRGRAYGL \u2506 0 \u2506 true \u2506 0.631656 \u2506 -649.034789 \u2506 -115.492025 \u2506 -114.952699 \u2506 -1.157456 \u2502\n", "\u2502 QAKWRLQTL \u2506 0 \u2506 false \u2506 0.7751346 \u2506 -355.210578 \u2506 -32.535676 \u2506 -31.298142 \u2506 -0.095934 \u2502\n", "\u2502 TPRVTGGGAM \u2506 0 \u2506 false \u2506 0.8708996 \u2506 -390.091896 \u2506 -46.326726 \u2506 -44.41779 \u2506 0.877601 \u2502\n", "\u2514\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2518" ] }, "execution_count": 4, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Every read-out of the frozen hold-out model, in one call.\n", "from tcren.score import CHANNELS, holdout_model, holdout_manifest, score_table\n", "\n", "model = holdout_model()\n", "scores = score_table(feats, iptm=feats[\"iptm\"].to_numpy(), model=model)\n", "panel = scores.with_columns(\n", " feats[\"epitope\"], feats[\"mhc\"], feats[\"y\"],\n", " feats[\"native_structure_exists\"].alias(\"covered\"),\n", " feats[\"iptm\"], feats[\"plddt\"])\n", "\n", "print(f\"model: tcren {model.tcren_version}, {model.n_pos} binders / {model.n_neg} non-binders \"\n", " f\"over {model.n_epitopes} hold-out epitopes\")\n", "print(f\"scored columns: {[c for c in scores.columns if c != 'complex.id']}\")\n", "panel.select(\"epitope\", \"y\", \"covered\", \"iptm\", \"pose_score\", \"binder_score\",\n", " \"binder_iptm\", \"confidence_residual\").head(5)\n" ] }, { "cell_type": "markdown", "id": "418c4343", "metadata": { "language": "markdown" }, "source": [ "### What the model was fitted on, and why this panel is a hold-out\n", "\n", "The frozen coefficients are reproducible rather than merely stated: `holdout_manifest()` returns the\n", "structures the fit used, with their dataset, epitope, label and ipTM, and `tcren fit-holdout`\n", "regenerates the shipped arrays from them.\n", "\n", "Two facts about the overlap, and they are different facts. **No structure in this panel contributed to\n", "the fit** \u2014 the intersection of the two id sets is empty. **Eleven of the panel's 22 epitopes do appear\n", "in the fit**, under other receptors; the model has seen those antigens, though never these complexes.\n", "The template-free stratum is where that matters least and where the honest reading is." ] }, { "cell_type": "code", "execution_count": 5, "id": "b4387fa4", "metadata": { "execution": { "iopub.execute_input": "2026-09-02T03:00:32.670621Z", "iopub.status.busy": "2026-09-02T03:00:32.670543Z", "iopub.status.idle": "2026-09-02T03:00:32.680583Z", "shell.execute_reply": "2026-09-02T03:00:32.680138Z" }, "language": "python" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "shape: (4, 2)\n", "\u250c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2510\n", "\u2502 dataset \u2506 len \u2502\n", "\u2502 --- \u2506 --- \u2502\n", "\u2502 str \u2506 u32 \u2502\n", "\u255e\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2561\n", "\u2502 VDJdb models \u2014 out-of-panel \u2506 6807 \u2502\n", "\u2502 IMMREP23 negatives \u2506 832 \u2502\n", "\u2502 VDJdb free pool \u2014 re-paired \u2506 431 \u2502\n", "\u2502 VDJdb free pool \u2014 cognate \u2506 222 \u2502\n", "\u2514\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2518\n", "\n", "structures shared with the fit 0 of 1089\n", "epitopes shared with the fit 11 of 22\n", " ATDALMTGF, AVFDRKSDAK, FLYALALLL, GILGFVFTL, GLCTLVAML, LLAGIGTVPI, LLWNGPMAV, NYNYLYRLF, RAKFKQLL, TPRVTGGGAM, YVLDHLIVV\n" ] } ], "source": [ "# The fit's provenance, and the two overlaps stated apart.\n", "man = holdout_manifest()\n", "shared_ids = set(feats[\"complex.id\"]) & set(man[\"pdb.id\"])\n", "shared_epitopes = sorted(set(feats[\"epitope\"]) & set(man[\"epitope\"]))\n", "\n", "print(man.group_by(\"dataset\").len().sort(\"len\", descending=True))\n", "print(f\"\\nstructures shared with the fit {len(shared_ids)} of {feats.height}\")\n", "print(f\"epitopes shared with the fit {len(shared_epitopes)} of {feats['epitope'].n_unique()}\")\n", "print(f\" {', '.join(shared_epitopes)}\")\n" ] }, { "cell_type": "markdown", "id": "5669d8ba", "metadata": { "language": "markdown" }, "source": [ "## 4 \u00b7 The ranking, one cohort at a time\n", "\n", "Each epitope cohort is its own ranking problem: the receptors are ranked within the cohort and scored\n", "against that cohort's own labels. Nothing is standardised across cohorts, and nothing is pooled.\n", "\n", "Three arms on identical rows, which is the reporting rule this benchmark follows: the generator's\n", "confidence alone, the structural read-out alone, and the two composed. A combination quoted without\n", "both of its parts cannot be checked." ] }, { "cell_type": "code", "execution_count": 6, "id": "a596b1a4", "metadata": { "execution": { "iopub.execute_input": "2026-09-02T03:00:32.681742Z", "iopub.status.busy": "2026-09-02T03:00:32.681672Z", "iopub.status.idle": "2026-09-02T03:00:32.926083Z", "shell.execute_reply": "2026-09-02T03:00:32.925728Z" }, "language": "python" }, "outputs": [ { "data": { "text/html": [ "
\n", "shape: (22, 8)
epitopecoverednn_posiptmbinder_scorebinder_iptmchannel_shape
strbooli64i64f64f64f64f64
"ELAGIGILTV"true65320.8759470.5409480.5495690.546336
"FLRGRAYGL"true52260.3727810.5630770.560.743077
"GILGFVFTL"true46230.947070.9316770.9316770.925466
"GLCTLVAML"true40200.61250.8027780.8027780.636111
"NYNYLYRLF"true58270.6236560.6785710.6771980.652473
"SLYNTVATL"true36140.7175320.6857140.6892860.678571
"ATDALMTGF"false46230.7051040.7278910.7301590.650794
"AVFDRKSDAK"false50250.35520.4143220.4143220.409207
"CLGGLLTMV"false46230.3591680.5576560.5557660.478261
"FLYALALLL"false69350.5218490.7878790.7869610.606979
"IVTDFSVIK"false60300.4955560.4457220.4469990.592593
"LLAGIGTVPI"false40200.78250.8654970.8625730.856725
"LLWNGPMAV"false51260.9569230.8317310.8349360.820513
"LLYDANYFL"false43190.8114040.4436090.4461150.631579
"PTDNYITTY"false54270.4897120.720.7184620.704615
"QAKWRLQTL"false44220.5289260.3484850.3484850.363636
"RAKFKQLL"false48240.5468750.7807970.7807970.695652
"RAQAPPPSW"false44200.0479170.5972220.5740740.703704
"RPIIRPATL"false43100.6333330.3290320.3290320.641935
"TPRVTGGGAM"false46230.6672970.633540.631470.538302
"TTDPSFLGRY"false62310.4984390.5055560.5055560.58
"YVLDHLIVV"false46230.4839320.6750.6781250.8
" ], "text/plain": [ "shape: (22, 8)\n", "\u250c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2510\n", "\u2502 epitope \u2506 covered \u2506 n \u2506 n_pos \u2506 iptm \u2506 binder_score \u2506 binder_iptm \u2506 channel_shape \u2502\n", "\u2502 --- \u2506 --- \u2506 --- \u2506 --- \u2506 --- \u2506 --- \u2506 --- \u2506 --- \u2502\n", "\u2502 str \u2506 bool \u2506 i64 \u2506 i64 \u2506 f64 \u2506 f64 \u2506 f64 \u2506 f64 \u2502\n", "\u255e\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2561\n", "\u2502 ELAGIGILTV \u2506 true \u2506 65 \u2506 32 \u2506 0.875947 \u2506 0.540948 \u2506 0.549569 \u2506 0.546336 \u2502\n", "\u2502 FLRGRAYGL \u2506 true \u2506 52 \u2506 26 \u2506 0.372781 \u2506 0.563077 \u2506 0.56 \u2506 0.743077 \u2502\n", "\u2502 GILGFVFTL \u2506 true \u2506 46 \u2506 23 \u2506 0.94707 \u2506 0.931677 \u2506 0.931677 \u2506 0.925466 \u2502\n", "\u2502 GLCTLVAML \u2506 true \u2506 40 \u2506 20 \u2506 0.6125 \u2506 0.802778 \u2506 0.802778 \u2506 0.636111 \u2502\n", "\u2502 NYNYLYRLF \u2506 true \u2506 58 \u2506 27 \u2506 0.623656 \u2506 0.678571 \u2506 0.677198 \u2506 0.652473 \u2502\n", "\u2502 SLYNTVATL \u2506 true \u2506 36 \u2506 14 \u2506 0.717532 \u2506 0.685714 \u2506 0.689286 \u2506 0.678571 \u2502\n", "\u2502 ATDALMTGF \u2506 false \u2506 46 \u2506 23 \u2506 0.705104 \u2506 0.727891 \u2506 0.730159 \u2506 0.650794 \u2502\n", "\u2502 AVFDRKSDAK \u2506 false \u2506 50 \u2506 25 \u2506 0.3552 \u2506 0.414322 \u2506 0.414322 \u2506 0.409207 \u2502\n", "\u2502 CLGGLLTMV \u2506 false \u2506 46 \u2506 23 \u2506 0.359168 \u2506 0.557656 \u2506 0.555766 \u2506 0.478261 \u2502\n", "\u2502 FLYALALLL \u2506 false \u2506 69 \u2506 35 \u2506 0.521849 \u2506 0.787879 \u2506 0.786961 \u2506 0.606979 \u2502\n", "\u2502 IVTDFSVIK \u2506 false \u2506 60 \u2506 30 \u2506 0.495556 \u2506 0.445722 \u2506 0.446999 \u2506 0.592593 \u2502\n", "\u2502 LLAGIGTVPI \u2506 false \u2506 40 \u2506 20 \u2506 0.7825 \u2506 0.865497 \u2506 0.862573 \u2506 0.856725 \u2502\n", "\u2502 LLWNGPMAV \u2506 false \u2506 51 \u2506 26 \u2506 0.956923 \u2506 0.831731 \u2506 0.834936 \u2506 0.820513 \u2502\n", "\u2502 LLYDANYFL \u2506 false \u2506 43 \u2506 19 \u2506 0.811404 \u2506 0.443609 \u2506 0.446115 \u2506 0.631579 \u2502\n", "\u2502 PTDNYITTY \u2506 false \u2506 54 \u2506 27 \u2506 0.489712 \u2506 0.72 \u2506 0.718462 \u2506 0.704615 \u2502\n", "\u2502 QAKWRLQTL \u2506 false \u2506 44 \u2506 22 \u2506 0.528926 \u2506 0.348485 \u2506 0.348485 \u2506 0.363636 \u2502\n", "\u2502 RAKFKQLL \u2506 false \u2506 48 \u2506 24 \u2506 0.546875 \u2506 0.780797 \u2506 0.780797 \u2506 0.695652 \u2502\n", "\u2502 RAQAPPPSW \u2506 false \u2506 44 \u2506 20 \u2506 0.047917 \u2506 0.597222 \u2506 0.574074 \u2506 0.703704 \u2502\n", "\u2502 RPIIRPATL \u2506 false \u2506 43 \u2506 10 \u2506 0.633333 \u2506 0.329032 \u2506 0.329032 \u2506 0.641935 \u2502\n", "\u2502 TPRVTGGGAM \u2506 false \u2506 46 \u2506 23 \u2506 0.667297 \u2506 0.63354 \u2506 0.63147 \u2506 0.538302 \u2502\n", "\u2502 TTDPSFLGRY \u2506 false \u2506 62 \u2506 31 \u2506 0.498439 \u2506 0.505556 \u2506 0.505556 \u2506 0.58 \u2502\n", "\u2502 YVLDHLIVV \u2506 false \u2506 46 \u2506 23 \u2506 0.483932 \u2506 0.675 \u2506 0.678125 \u2506 0.8 \u2502\n", "\u2514\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2518" ] }, "execution_count": 6, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Per-cohort ROC-AUC, one row per epitope. No pooling, no cross-cohort standardisation.\n", "from sklearn.metrics import average_precision_score, roc_auc_score\n", "\n", "ARMS = [\"iptm\", \"plddt\", \"pose_score\", \"binder_score\", \"binder_iptm\", \"confidence_residual\"] + [\n", " f\"channel_{c}\" for c in CHANNELS if f\"channel_{c}\" in panel.columns]\n", "\n", "\n", "def per_cohort(table, arms=ARMS, metric=roc_auc_score):\n", " \"\"\"One row per epitope: n, positives, the template flag, and `metric` for every arm.\"\"\"\n", " out = []\n", " for epitope in sorted(set(table[\"epitope\"])):\n", " g = table.filter(pl.col(\"epitope\") == epitope)\n", " y = g[\"y\"].to_numpy()\n", " row = {\"epitope\": epitope, \"covered\": bool(g[\"covered\"][0]),\n", " \"n\": len(y), \"n_pos\": int(y.sum())}\n", " for a in arms:\n", " v = g[a].to_numpy().astype(float)\n", " ok = np.isfinite(v)\n", " row[a] = metric(y[ok], v[ok]) if ok.sum() > 2 and len(set(y[ok])) == 2 else np.nan\n", " out.append(row)\n", " return pl.DataFrame(out).sort([\"covered\", \"epitope\"], descending=[True, False])\n", "\n", "\n", "roc = per_cohort(panel)\n", "pl.Config.set_tbl_rows(30)\n", "roc.select(\"epitope\", \"covered\", \"n\", \"n_pos\", \"iptm\", \"binder_score\", \"binder_iptm\",\n", " \"channel_shape\")\n" ] }, { "cell_type": "markdown", "id": "afa08e8b", "metadata": { "language": "markdown" }, "source": [ "### The two strata, reported apart\n", "\n", "The median over cohorts and the count of cohorts beating no-skill, per stratum. The clearance count is\n", "the part a macro mean throws away: it says how many of the individual ranking problems the arm actually\n", "solves, and the target on this panel is more than 60 % cleared. For ROC-AUC no-skill is 0.50; the same\n", "helper is reused below for average precision, where no-skill is the cohort's own prevalence instead.\n" ] }, { "cell_type": "code", "execution_count": 7, "id": "72d2a678", "metadata": { "execution": { "iopub.execute_input": "2026-09-02T03:00:32.928031Z", "iopub.status.busy": "2026-09-02T03:00:32.927900Z", "iopub.status.idle": "2026-09-02T03:00:32.933159Z", "shell.execute_reply": "2026-09-02T03:00:32.932770Z" }, "language": "python" }, "outputs": [ { "data": { "text/html": [ "
\n", "shape: (22, 5)
stratumcohortsarmmediancleared
stri64strf64i64
"template-covered"6"iptm"0.6705945
"template-covered"6"plddt"0.7312125
"template-covered"6"pose_score"0.7134625
"template-covered"6"binder_score"0.6821436
"template-covered"6"binder_iptm"0.6832426
"template-covered"6"confidence_residual"0.5943245
"template-covered"6"channel_placement"0.6247614
"template-covered"6"channel_interface"0.6730716
"template-covered"6"channel_shape"0.6655226
"template-covered"6"channel_energetics"0.6916216
"template-covered"6"channel_mechanics"0.6991615
"template-free"16"iptm"0.5253879
"template-free"16"plddt"0.5861711
"template-free"16"pose_score"0.61243111
"template-free"16"binder_score"0.61538111
"template-free"16"binder_iptm"0.60277211
"template-free"16"confidence_residual"0.4953297
"template-free"16"channel_placement"0.5046588
"template-free"16"channel_interface"0.499828
"template-free"16"channel_shape"0.63675713
"template-free"16"channel_energetics"0.58120611
"template-free"16"channel_mechanics"0.52524110
" ], "text/plain": [ "shape: (22, 5)\n", "\u250c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2510\n", "\u2502 stratum \u2506 cohorts \u2506 arm \u2506 median \u2506 cleared \u2502\n", "\u2502 --- \u2506 --- \u2506 --- \u2506 --- \u2506 --- \u2502\n", "\u2502 str \u2506 i64 \u2506 str \u2506 f64 \u2506 i64 \u2502\n", "\u255e\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2561\n", "\u2502 template-covered \u2506 6 \u2506 iptm \u2506 0.670594 \u2506 5 \u2502\n", "\u2502 template-covered \u2506 6 \u2506 plddt \u2506 0.731212 \u2506 5 \u2502\n", "\u2502 template-covered \u2506 6 \u2506 pose_score \u2506 0.713462 \u2506 5 \u2502\n", "\u2502 template-covered \u2506 6 \u2506 binder_score \u2506 0.682143 \u2506 6 \u2502\n", "\u2502 template-covered \u2506 6 \u2506 binder_iptm \u2506 0.683242 \u2506 6 \u2502\n", "\u2502 template-covered \u2506 6 \u2506 confidence_residual \u2506 0.594324 \u2506 5 \u2502\n", "\u2502 template-covered \u2506 6 \u2506 channel_placement \u2506 0.624761 \u2506 4 \u2502\n", "\u2502 template-covered \u2506 6 \u2506 channel_interface \u2506 0.673071 \u2506 6 \u2502\n", "\u2502 template-covered \u2506 6 \u2506 channel_shape \u2506 0.665522 \u2506 6 \u2502\n", "\u2502 template-covered \u2506 6 \u2506 channel_energetics \u2506 0.691621 \u2506 6 \u2502\n", "\u2502 template-covered \u2506 6 \u2506 channel_mechanics \u2506 0.699161 \u2506 5 \u2502\n", "\u2502 template-free \u2506 16 \u2506 iptm \u2506 0.525387 \u2506 9 \u2502\n", "\u2502 template-free \u2506 16 \u2506 plddt \u2506 0.58617 \u2506 11 \u2502\n", "\u2502 template-free \u2506 16 \u2506 pose_score \u2506 0.612431 \u2506 11 \u2502\n", "\u2502 template-free \u2506 16 \u2506 binder_score \u2506 0.615381 \u2506 11 \u2502\n", "\u2502 template-free \u2506 16 \u2506 binder_iptm \u2506 0.602772 \u2506 11 \u2502\n", "\u2502 template-free \u2506 16 \u2506 confidence_residual \u2506 0.495329 \u2506 7 \u2502\n", "\u2502 template-free \u2506 16 \u2506 channel_placement \u2506 0.504658 \u2506 8 \u2502\n", "\u2502 template-free \u2506 16 \u2506 channel_interface \u2506 0.49982 \u2506 8 \u2502\n", "\u2502 template-free \u2506 16 \u2506 channel_shape \u2506 0.636757 \u2506 13 \u2502\n", "\u2502 template-free \u2506 16 \u2506 channel_energetics \u2506 0.581206 \u2506 11 \u2502\n", "\u2502 template-free \u2506 16 \u2506 channel_mechanics \u2506 0.525241 \u2506 10 \u2502\n", "\u2514\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2518" ] }, "execution_count": 7, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Median per-cohort score and cohorts cleared, template-covered against template-free.\n", "# \"Cleared\" is judged against the arm's own no-skill value: 0.50 for ROC-AUC, and the cohort's\n", "# own positive prevalence for average precision, which is not 0.50 and differs cohort to cohort.\n", "def stratum_summary(table, arms=ARMS, baseline=\"roc\"):\n", " \"\"\"Per stratum and arm: the median over cohorts, and how many cohorts beat no-skill.\"\"\"\n", " ref = (np.full(table.height, 0.5) if baseline == \"roc\"\n", " else table[\"n_pos\"].to_numpy() / table[\"n\"].to_numpy())\n", " table = table.with_columns(pl.Series(\"no_skill\", ref))\n", " out = []\n", " for covered in (True, False):\n", " sub = table.filter(pl.col(\"covered\") == covered)\n", " for a in arms:\n", " v = sub[a].to_numpy()\n", " out.append({\"stratum\": \"template-covered\" if covered else \"template-free\",\n", " \"cohorts\": sub.height, \"arm\": a,\n", " \"median\": float(np.nanmedian(v)),\n", " \"cleared\": int(np.nansum(v > sub[\"no_skill\"].to_numpy()))})\n", " return pl.DataFrame(out)\n", "\n", "\n", "summary = stratum_summary(roc)\n", "pl.Config.set_tbl_rows(40)\n", "summary\n" ] }, { "cell_type": "markdown", "id": "50430db1", "metadata": { "language": "markdown" }, "source": [ "Two readings, and they are the result this benchmark exists for.\n", "\n", "**Where a template exists, the structure and the confidence carry different evidence and compose.** The\n", "posterior clears 0.50 in every covered cohort, and so does the composition with ipTM.\n", "\n", "**Where no template exists, the generator's confidence falls to near a coin and does not warn you.**\n", "The structural read-outs do not fall as far: the posterior and the one-class pose distance both stay\n", "above ipTM's median, and the **shape channel is the strongest single arm on this stratum** \u2014 the\n", "footprint's shape is invariant under the rigid-body placement the co-folding model is optimising,\n", "which is why it still says something once that model has produced a confident pose." ] }, { "cell_type": "markdown", "id": "4db5aed1", "metadata": { "language": "markdown" }, "source": [ "## 5 \u00b7 Which part of the structure says so \u2014 the five channels\n", "\n", "A marginal of a Gaussian is a sub-block of its covariance: exact, closed form, no re-fit. So asking\n", "which family of descriptors carries the signal costs an index and nothing else. The five channels are\n", "`placement` (where the receptor sits in the groove frame), `interface` (how much interface it makes, of\n", "what chemistry), `shape` (the footprint free of its size), `energetics` (the contact chemistry in kT)\n", "and `mechanics` (the interface as a network of breakable springs).\n", "\n", "They do not sum to `binder_score` and should not: the whole model also reads the correlations *between*\n", "channels, which a per-channel view cannot show. What they give is attribution \u2014 and sometimes a better\n", "instrument, where the whole model dilutes a channel that is carrying the cohort on its own." ] }, { "cell_type": "code", "execution_count": 8, "id": "0e150e19", "metadata": { "execution": { "iopub.execute_input": "2026-09-02T03:00:32.934367Z", "iopub.status.busy": "2026-09-02T03:00:32.934289Z", "iopub.status.idle": "2026-09-02T03:00:32.937626Z", "shell.execute_reply": "2026-09-02T03:00:32.937243Z" }, "language": "python" }, "outputs": [ { "data": { "text/html": [ "
\n", "shape: (22, 8)
epitopecoveredchannel_placementchannel_interfacechannel_shapechannel_energeticschannel_mechanicsbinder_score
strboolf64f64f64f64f64f64
"ELAGIGILTV"true0.5689660.6950430.5463360.7101290.5808190.540948
"FLRGRAYGL"true0.2323080.7061540.7430770.6184620.4476920.563077
"GILGFVFTL"true0.7784680.7846790.9254660.9109730.8074530.931677
"GLCTLVAML"true0.6805560.6111110.6361110.6222220.7472220.802778
"NYNYLYRLF"true0.7211540.6510990.6524730.704670.6510990.678571
"SLYNTVATL"true0.4964290.6035710.6785710.6785710.8892860.685714
"ATDALMTGF"false0.6507940.532880.6507940.5034010.6099770.727891
"AVFDRKSDAK"false0.4782610.4833760.4092070.6035810.5268540.414322
"CLGGLLTMV"false0.5992440.4158790.4782610.5860110.4990550.557656
"FLYALALLL"false0.8677690.5491280.6069790.7741050.441690.787879
"IVTDFSVIK"false0.3780330.4457220.5925930.5772670.5236270.445722
"LLAGIGTVPI"false0.50.9093570.8567250.8713450.698830.865497
"LLWNGPMAV"false0.7099360.6506410.8205130.6618590.7371790.831731
"LLYDANYFL"false0.2882210.4260650.6315790.3684210.6340850.443609
"PTDNYITTY"false0.6676920.3630770.7046150.5538460.5969230.72
"QAKWRLQTL"false0.3874460.4307360.3636360.3831170.43290.348485
"RAKFKQLL"false0.4800720.5199280.6956520.5851450.5054350.780797
"RAQAPPPSW"false0.3009260.7083330.7037040.2314810.620370.597222
"RPIIRPATL"false0.4161290.4903230.6419350.3580650.4064520.329032
"TPRVTGGGAM"false0.5093170.5093170.5383020.6935820.62940.63354
"TTDPSFLGRY"false0.6333330.5711110.580.4588890.3711110.505556
"YVLDHLIVV"false0.6031250.4281250.80.6250.4468750.675
" ], "text/plain": [ "shape: (22, 8)\n", "\u250c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2510\n", "\u2502 epitope \u2506 covered \u2506 channel_pl \u2506 channel_in \u2506 channel_sh \u2506 channel_en \u2506 channel_m \u2506 binder_sc \u2502\n", "\u2502 --- \u2506 --- \u2506 acement \u2506 terface \u2506 ape \u2506 ergetics \u2506 echanics \u2506 ore \u2502\n", "\u2502 str \u2506 bool \u2506 --- \u2506 --- \u2506 --- \u2506 --- \u2506 --- \u2506 --- \u2502\n", "\u2502 \u2506 \u2506 f64 \u2506 f64 \u2506 f64 \u2506 f64 \u2506 f64 \u2506 f64 \u2502\n", "\u255e\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2561\n", "\u2502 ELAGIGILTV \u2506 true \u2506 0.568966 \u2506 0.695043 \u2506 0.546336 \u2506 0.710129 \u2506 0.580819 \u2506 0.540948 \u2502\n", "\u2502 FLRGRAYGL \u2506 true \u2506 0.232308 \u2506 0.706154 \u2506 0.743077 \u2506 0.618462 \u2506 0.447692 \u2506 0.563077 \u2502\n", "\u2502 GILGFVFTL \u2506 true \u2506 0.778468 \u2506 0.784679 \u2506 0.925466 \u2506 0.910973 \u2506 0.807453 \u2506 0.931677 \u2502\n", "\u2502 GLCTLVAML \u2506 true \u2506 0.680556 \u2506 0.611111 \u2506 0.636111 \u2506 0.622222 \u2506 0.747222 \u2506 0.802778 \u2502\n", "\u2502 NYNYLYRLF \u2506 true \u2506 0.721154 \u2506 0.651099 \u2506 0.652473 \u2506 0.70467 \u2506 0.651099 \u2506 0.678571 \u2502\n", "\u2502 SLYNTVATL \u2506 true \u2506 0.496429 \u2506 0.603571 \u2506 0.678571 \u2506 0.678571 \u2506 0.889286 \u2506 0.685714 \u2502\n", "\u2502 ATDALMTGF \u2506 false \u2506 0.650794 \u2506 0.53288 \u2506 0.650794 \u2506 0.503401 \u2506 0.609977 \u2506 0.727891 \u2502\n", "\u2502 AVFDRKSDAK \u2506 false \u2506 0.478261 \u2506 0.483376 \u2506 0.409207 \u2506 0.603581 \u2506 0.526854 \u2506 0.414322 \u2502\n", "\u2502 CLGGLLTMV \u2506 false \u2506 0.599244 \u2506 0.415879 \u2506 0.478261 \u2506 0.586011 \u2506 0.499055 \u2506 0.557656 \u2502\n", "\u2502 FLYALALLL \u2506 false \u2506 0.867769 \u2506 0.549128 \u2506 0.606979 \u2506 0.774105 \u2506 0.44169 \u2506 0.787879 \u2502\n", "\u2502 IVTDFSVIK \u2506 false \u2506 0.378033 \u2506 0.445722 \u2506 0.592593 \u2506 0.577267 \u2506 0.523627 \u2506 0.445722 \u2502\n", "\u2502 LLAGIGTVPI \u2506 false \u2506 0.5 \u2506 0.909357 \u2506 0.856725 \u2506 0.871345 \u2506 0.69883 \u2506 0.865497 \u2502\n", "\u2502 LLWNGPMAV \u2506 false \u2506 0.709936 \u2506 0.650641 \u2506 0.820513 \u2506 0.661859 \u2506 0.737179 \u2506 0.831731 \u2502\n", "\u2502 LLYDANYFL \u2506 false \u2506 0.288221 \u2506 0.426065 \u2506 0.631579 \u2506 0.368421 \u2506 0.634085 \u2506 0.443609 \u2502\n", "\u2502 PTDNYITTY \u2506 false \u2506 0.667692 \u2506 0.363077 \u2506 0.704615 \u2506 0.553846 \u2506 0.596923 \u2506 0.72 \u2502\n", "\u2502 QAKWRLQTL \u2506 false \u2506 0.387446 \u2506 0.430736 \u2506 0.363636 \u2506 0.383117 \u2506 0.4329 \u2506 0.348485 \u2502\n", "\u2502 RAKFKQLL \u2506 false \u2506 0.480072 \u2506 0.519928 \u2506 0.695652 \u2506 0.585145 \u2506 0.505435 \u2506 0.780797 \u2502\n", "\u2502 RAQAPPPSW \u2506 false \u2506 0.300926 \u2506 0.708333 \u2506 0.703704 \u2506 0.231481 \u2506 0.62037 \u2506 0.597222 \u2502\n", "\u2502 RPIIRPATL \u2506 false \u2506 0.416129 \u2506 0.490323 \u2506 0.641935 \u2506 0.358065 \u2506 0.406452 \u2506 0.329032 \u2502\n", "\u2502 TPRVTGGGAM \u2506 false \u2506 0.509317 \u2506 0.509317 \u2506 0.538302 \u2506 0.693582 \u2506 0.6294 \u2506 0.63354 \u2502\n", "\u2502 TTDPSFLGRY \u2506 false \u2506 0.633333 \u2506 0.571111 \u2506 0.58 \u2506 0.458889 \u2506 0.371111 \u2506 0.505556 \u2502\n", "\u2502 YVLDHLIVV \u2506 false \u2506 0.603125 \u2506 0.428125 \u2506 0.8 \u2506 0.625 \u2506 0.446875 \u2506 0.675 \u2502\n", "\u2514\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2518" ] }, "execution_count": 8, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# The five channels per cohort, and their per-stratum medians beside the full posterior.\n", "channel_cols = [f\"channel_{c}\" for c in CHANNELS if f\"channel_{c}\" in panel.columns]\n", "roc.select([\"epitope\", \"covered\", *channel_cols, \"binder_score\"])\n" ] }, { "cell_type": "code", "execution_count": 9, "id": "7ffd77eb", "metadata": { "execution": { "iopub.execute_input": "2026-09-02T03:00:32.938630Z", "iopub.status.busy": "2026-09-02T03:00:32.938551Z", "iopub.status.idle": "2026-09-02T03:00:32.941627Z", "shell.execute_reply": "2026-09-02T03:00:32.941340Z" }, "language": "python" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "template-covered (6 cohorts)\n", " channel_mechanics median ROC-AUC 0.699 cleared 5/6\n", " channel_energetics median ROC-AUC 0.692 cleared 6/6\n", " binder_score median ROC-AUC 0.682 cleared 6/6\n", " channel_interface median ROC-AUC 0.673 cleared 6/6\n", " iptm median ROC-AUC 0.671 cleared 5/6\n", " channel_shape median ROC-AUC 0.666 cleared 6/6\n", " channel_placement median ROC-AUC 0.625 cleared 4/6\n", "\n", "template-free (16 cohorts)\n", " channel_shape median ROC-AUC 0.637 cleared 13/16\n", " binder_score median ROC-AUC 0.615 cleared 11/16\n", " channel_energetics median ROC-AUC 0.581 cleared 11/16\n", " iptm median ROC-AUC 0.525 cleared 9/16\n", " channel_mechanics median ROC-AUC 0.525 cleared 10/16\n", " channel_placement median ROC-AUC 0.505 cleared 8/16\n", " channel_interface median ROC-AUC 0.500 cleared 8/16\n", "\n" ] } ], "source": [ "# Channel medians per stratum, sorted, against the full posterior on the same rows.\n", "for covered in (True, False):\n", " sub = roc.filter(pl.col(\"covered\") == covered)\n", " label = \"template-covered\" if covered else \"template-free\"\n", " print(f\"{label} ({sub.height} cohorts)\")\n", " ranked = sorted(((float(np.nanmedian(sub[c].to_numpy())), c)\n", " for c in [*channel_cols, \"binder_score\", \"iptm\"]), reverse=True)\n", " for value, name in ranked:\n", " cleared = int(np.nansum(sub[name].to_numpy() > 0.5))\n", " print(f\" {name:22s} median ROC-AUC {value:.3f} cleared {cleared}/{sub.height}\")\n", " print()\n" ] }, { "cell_type": "markdown", "id": "f60b461c", "metadata": { "language": "markdown" }, "source": [ "## 6 \u00b7 The panel, drawn\n", "\n", "One point per cohort per arm, with the 0.50 line marked. The two strata are drawn on separate axes\n", "because they are separate claims." ] }, { "cell_type": "code", "execution_count": 10, "id": "77056709", "metadata": { "execution": { "iopub.execute_input": "2026-09-02T03:00:32.942794Z", "iopub.status.busy": "2026-09-02T03:00:32.942718Z", "iopub.status.idle": "2026-09-02T03:00:33.387679Z", "shell.execute_reply": "2026-09-02T03:00:33.387329Z" }, "language": "python" }, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "execution_count": 10, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "%matplotlib inline\n", "# Per-cohort ROC-AUC by stratum: one point per cohort, 0.50 marked. Okabe-Ito palette.\n", "import matplotlib.pyplot as plt\n", "\n", "PLOT_ARMS = [\"iptm\", \"plddt\", \"binder_score\", \"binder_iptm\", \"channel_shape\", \"pose_score\"]\n", "OKABE = {\"iptm\": \"#56B4E9\", \"plddt\": \"#0072B2\", \"binder_score\": \"#D55E00\",\n", " \"binder_iptm\": \"#000000\", \"channel_shape\": \"#E69F00\", \"pose_score\": \"#009E73\"}\n", "\n", "plt.rcParams.update({\"font.size\": 8, \"axes.spines.top\": False, \"axes.spines.right\": False,\n", " \"figure.dpi\": 130})\n", "fig, axes = plt.subplots(1, 2, figsize=(9.0, 3.4), sharey=True,\n", " gridspec_kw={\"width_ratios\": [6, 16], \"wspace\": 0.06})\n", "\n", "for ax, covered in zip(axes, (True, False)):\n", " sub = roc.filter(pl.col(\"covered\") == covered)\n", " x = np.arange(sub.height)\n", " for k, arm in enumerate(PLOT_ARMS):\n", " ax.scatter(x + (k - 2.5) * 0.11, sub[arm].to_numpy(), s=16, color=OKABE[arm],\n", " label=arm if covered else None, zorder=3, edgecolor=\"none\")\n", " ax.axhline(0.5, color=\"0.6\", lw=0.8, zorder=1)\n", " ax.set_xticks(x)\n", " ax.set_xticklabels(sub[\"epitope\"].to_list(), rotation=90)\n", " ax.set_title(f\"{'template-covered' if covered else 'template-free'} \"\n", " f\"({sub.height} cohorts, n = {int(sub['n'].sum())})\", fontsize=8)\n", " ax.set_ylim(0.0, 1.02)\n", "\n", "axes[0].set_ylabel(\"ROC-AUC within cohort\")\n", "fig.legend(loc=\"lower center\", ncol=6, frameon=False, bbox_to_anchor=(0.5, -0.30))\n", "fig\n" ] }, { "cell_type": "markdown", "id": "728b117d", "metadata": { "language": "markdown" }, "source": [ "## 7 \u00b7 Precision, and where the ranking is used\n", "\n", "ROC-AUC is the right read-out when the whole ordering matters. When only the head of the list will be\n", "tested, average precision is the one to quote, and both are reported here on identical rows so neither\n", "has to be taken on trust." ] }, { "cell_type": "code", "execution_count": 11, "id": "c70b22ab", "metadata": { "execution": { "iopub.execute_input": "2026-09-02T03:00:33.389001Z", "iopub.status.busy": "2026-09-02T03:00:33.388860Z", "iopub.status.idle": "2026-09-02T03:00:33.479550Z", "shell.execute_reply": "2026-09-02T03:00:33.479085Z" }, "language": "python" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "cohort prevalence: 0.233 to 0.510\n" ] }, { "data": { "text/html": [ "
\n", "shape: (8, 5)
stratumcohortsarmmediancleared
stri64strf64i64
"template-covered"6"iptm"0.6601675
"template-covered"6"binder_score"0.6329436
"template-covered"6"binder_iptm"0.6360796
"template-covered"6"channel_shape"0.6507746
"template-free"16"iptm"0.55320511
"template-free"16"binder_score"0.60886711
"template-free"16"binder_iptm"0.60776811
"template-free"16"channel_shape"0.64855613
" ], "text/plain": [ "shape: (8, 5)\n", "\u250c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u252c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2510\n", "\u2502 stratum \u2506 cohorts \u2506 arm \u2506 median \u2506 cleared \u2502\n", "\u2502 --- \u2506 --- \u2506 --- \u2506 --- \u2506 --- \u2502\n", "\u2502 str \u2506 i64 \u2506 str \u2506 f64 \u2506 i64 \u2502\n", "\u255e\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u256a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2561\n", "\u2502 template-covered \u2506 6 \u2506 iptm \u2506 0.660167 \u2506 5 \u2502\n", "\u2502 template-covered \u2506 6 \u2506 binder_score \u2506 0.632943 \u2506 6 \u2502\n", "\u2502 template-covered \u2506 6 \u2506 binder_iptm \u2506 0.636079 \u2506 6 \u2502\n", "\u2502 template-covered \u2506 6 \u2506 channel_shape \u2506 0.650774 \u2506 6 \u2502\n", "\u2502 template-free \u2506 16 \u2506 iptm \u2506 0.553205 \u2506 11 \u2502\n", "\u2502 template-free \u2506 16 \u2506 binder_score \u2506 0.608867 \u2506 11 \u2502\n", "\u2502 template-free \u2506 16 \u2506 binder_iptm \u2506 0.607768 \u2506 11 \u2502\n", "\u2502 template-free \u2506 16 \u2506 channel_shape \u2506 0.648556 \u2506 13 \u2502\n", "\u2514\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2534\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2518" ] }, "execution_count": 11, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# The same cohorts under average precision, so the head-of-list behaviour is visible too.\n", "# Clearance is against each cohort's own prevalence here, never against 0.50.\n", "pr = per_cohort(panel, metric=average_precision_score)\n", "pr_summary = stratum_summary(pr, baseline=\"prevalence\")\n", "print(\"cohort prevalence: \"\n", " f\"{(roc['n_pos'] / roc['n']).min():.3f} to {(roc['n_pos'] / roc['n']).max():.3f}\")\n", "pr_summary.filter(pl.col(\"arm\").is_in([\"iptm\", \"binder_score\", \"binder_iptm\", \"channel_shape\"]))\n" ] }, { "cell_type": "markdown", "id": "a72c24e5", "metadata": { "language": "markdown" }, "source": [ "## What this notebook establishes\n", "\n", "- The whole path runs from deposited coordinates: fetch, `tcren features`, `tcren.score.score_table`.\n", " No binding assay is read at any point, and no structure scored here entered the frozen fit.\n", "- The panel is reported **case by case over 22 cohorts**, template-covered and template-free apart. A\n", " macro average over the two strata is a Simpson artefact on this benchmark and is not quoted.\n", "- The five channels attribute a score to a part of the structure, at the cost of one index into a\n", " covariance, and on the template-free stratum the shape channel is the strongest single arm.\n", "\n", "**What it does not claim.** Not affinity: the score set ranks specificity, and what a static interface\n", "reads of dynamics is the off-rate rather than the equilibrium constant. Not a substitute for reporting\n", "template coverage: nothing computable from the model announces which regime a cohort is in \u2014 the\n", "generator's confidence least of all \u2014 so template availability is a covariate to report, never one to\n", "infer.\n", "\n", "See `docs/assess.rst` for the read-outs one at a time, and `rank_peptides_cpl.ipynb` for the other\n", "ranking task: peptides against a fixed receptor." ] } ], "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": "3.12.13" } }, "nbformat": 4, "nbformat_minor": 5 }