{ "cells": [ { "cell_type": "markdown", "metadata": { "id": "k9qe_aUJSeho" }, "source": [ "# Knowledge-primed Neural Networks (KPNNs) for single cell data\n", "\n", "In this tutorial we will show how CORNETO can be used to build custom neural network architectures informed by prior knowledge. We will see how to implement a knowledge-primed neural network1. We will use the single cell data from the publication \"Knowledge-primed neural networks enable biologically interpretable deep learning on single-cell sequencing data\", from Nikolaus Fortelny & Christoph Bock, where they used a single-cell RNA-seq dataset they previously generated2, which measures cellular responses to T cell receptor (TCR) stimulation in a standardized in vitro model. The dataset was chosen due to the TCR signaling pathway's complexity and its well-characterized role in orchestrating transcriptional responses to antigen detection in T cells.\n", "\n", "## Why CORNETO?\n", "\n", "In the original publication, authors built a KPNN by searching on databases, building a Direct Acyclic Graph (DAG) by running shortest paths from TCR receptor to genes. However, this approach is not optimal. CORNETO, thanks to its advanced capabilities for modeling and optimization on networks, provides methods to automatically find DAG architectures in an optimal way. \n", "\n", "In addition to this, CORNETO provides methods to build DAG NN architectures with ease using Keras +3, making KPNN implementation very flexible and interoperable with backends like Pytorch, Tensorflow and JAX.\n", "\n", "## How does it work?\n", "\n", "Thanks to CORNETO's building blocks for optimization over networks, we can easily model optimization problems to find DAG architectures from a Prior Knowledge Network. After we have the backbone, we can convert it to a neural network using the utility functions included in CORNETO.\n", "\n", "\n", "## References\n", "1. Fortelny, N., & Bock, C. (2020). Knowledge-primed neural networks enable biologically interpretable deep learning on single-cell sequencing data. Genome biology, 21, 1-36.\n", "2. Datlinger, P., Rendeiro, A. F., Schmidl, C., Krausgruber, T., Traxler, P., Klughammer, J., ... & Bock, C. (2017). Pooled CRISPR screening with single-cell transcriptome readout. Nature methods, 14(3), 297-301.\n", "\n" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Download and import the single cell dataset" ] }, { "cell_type": "code", "execution_count": 1, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "AYKxvyAC_ZCn", "outputId": "c4c55b18-5438-4f94-de96-bd402e760dc6" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Effective URL: https://medical-epigenomics.org/papers/fortelny2019/\n", "Downloading https://medical-epigenomics.org/papers/fortelny2019/TCR_Edgelist.csv to /var/folders/b4/gwkwsdb93sv11rtztqbm3l040000gn/T/tmpb21ulczk/TCR_Edgelist.csv\n", "Downloading https://medical-epigenomics.org/papers/fortelny2019/TCR_ClassLabels.csv to /var/folders/b4/gwkwsdb93sv11rtztqbm3l040000gn/T/tmpb21ulczk/TCR_ClassLabels.csv\n", "Downloading https://medical-epigenomics.org/papers/fortelny2019/TCR_Data.h5 to /var/folders/b4/gwkwsdb93sv11rtztqbm3l040000gn/T/tmpb21ulczk/TCR_Data.h5\n", "Downloaded files:\n", "/var/folders/b4/gwkwsdb93sv11rtztqbm3l040000gn/T/tmpb21ulczk/TCR_Edgelist.csv\n", "/var/folders/b4/gwkwsdb93sv11rtztqbm3l040000gn/T/tmpb21ulczk/TCR_ClassLabels.csv\n", "/var/folders/b4/gwkwsdb93sv11rtztqbm3l040000gn/T/tmpb21ulczk/TCR_Data.h5\n" ] } ], "source": [ "import os\n", "import urllib.request\n", "import urllib.parse\n", "import tempfile\n", "import pandas as pd\n", "import scanpy as sc\n", "import numpy as np\n", "import corneto as cn\n", "\n", "with urllib.request.urlopen(\"http://kpnn.computational-epigenetics.org/\") as response:\n", " web_input = response.geturl()\n", "print(\"Effective URL:\", web_input)\n", "\n", "files = [\"TCR_Edgelist.csv\", \"TCR_ClassLabels.csv\", \"TCR_Data.h5\"]\n", "\n", "temp_dir = tempfile.mkdtemp()\n", "\n", "# Download files\n", "file_paths = []\n", "for file in files:\n", " url = urllib.parse.urljoin(web_input, file)\n", " output_path = os.path.join(temp_dir, file)\n", " print(f\"Downloading {url} to {output_path}\")\n", " try:\n", " with urllib.request.urlopen(url) as response:\n", " with open(output_path, 'wb') as f:\n", " f.write(response.read())\n", " file_paths.append(output_path)\n", " except Exception as e:\n", " print(f\"Failed to download {url}: {e}\")\n", "\n", "print(\"Downloaded files:\")\n", "for path in file_paths:\n", " print(path)" ] }, { "cell_type": "code", "execution_count": 2, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "9qaRPRLeA05k", "outputId": "21b2325e-5603-434d-a65e-996925754bb3" }, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "/Users/pablorodriguezmier/miniforge3/envs/corneto/lib/python3.12/site-packages/anndata/_core/anndata.py:1758: UserWarning: Variable names are not unique. To make them unique, call `.var_names_make_unique`.\n", " utils.warn_names_duplicates(\"var\")\n" ] } ], "source": [ "# The data contains also the original network they built with shortest paths.\n", "# We will use it to replicate the study\n", "df_edges = pd.read_csv(file_paths[0])\n", "df_labels = pd.read_csv(file_paths[1])\n", "# Import the 10x data with Scanpy\n", "adata = sc.read_10x_h5(file_paths[2])" ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "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", "
barcodeTCR
0AAACCTGCACACATGT-10
1AAACCTGCACGTCTCT-10
2AAACCTGTCAATACCG-10
3AAACCTGTCGTGGTCG-10
4AAACGGGTCTGAGTGT-10
.........
1730TTTCCTCGTCATGCCG-21
1731TTTGCGCGTAGCCTCG-21
1732TTTGGTTAGATACACA-21
1733TTTGGTTGTATGAATG-21
1734TTTGGTTTCCAAGTAC-21
\n", "

1735 rows × 2 columns

\n", "
" ], "text/plain": [ " barcode TCR\n", "0 AAACCTGCACACATGT-1 0\n", "1 AAACCTGCACGTCTCT-1 0\n", "2 AAACCTGTCAATACCG-1 0\n", "3 AAACCTGTCGTGGTCG-1 0\n", "4 AAACGGGTCTGAGTGT-1 0\n", "... ... ...\n", "1730 TTTCCTCGTCATGCCG-2 1\n", "1731 TTTGCGCGTAGCCTCG-2 1\n", "1732 TTTGGTTAGATACACA-2 1\n", "1733 TTTGGTTGTATGAATG-2 1\n", "1734 TTTGGTTTCCAAGTAC-2 1\n", "\n", "[1735 rows x 2 columns]" ] }, "execution_count": 3, "metadata": {}, "output_type": "execute_result" } ], "source": [ "df_labels" ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "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", "
parentchild
0TCRZAP70
1ZAP70MAPK14
2MAPK14FOXO3
3MAPK14STAT1
4MAPK14STAT3
.........
27574HMGA1MTRNR2L9_gene
27575MYBC12orf50_gene
27576MYBTRPC5OS_gene
27577SOX2TRPC5OS_gene
27578CRTC1MTRNR2L9_gene
\n", "

27579 rows × 2 columns

\n", "
" ], "text/plain": [ " parent child\n", "0 TCR ZAP70\n", "1 ZAP70 MAPK14\n", "2 MAPK14 FOXO3\n", "3 MAPK14 STAT1\n", "4 MAPK14 STAT3\n", "... ... ...\n", "27574 HMGA1 MTRNR2L9_gene\n", "27575 MYB C12orf50_gene\n", "27576 MYB TRPC5OS_gene\n", "27577 SOX2 TRPC5OS_gene\n", "27578 CRTC1 MTRNR2L9_gene\n", "\n", "[27579 rows x 2 columns]" ] }, "execution_count": 4, "metadata": {}, "output_type": "execute_result" } ], "source": [ "df_edges" ] }, { "cell_type": "code", "execution_count": 5, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 423 }, "id": "gZVpIe8UN011", "outputId": "6a468338-26e6-4ac4-b115-af6169126445" }, "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", "
gene_ids
DDX11L1ENSG00000223972
WASH7PENSG00000227232
MIR6859-2ENSG00000278267
MIR1302-10ENSG00000243485
MIR1302-11ENSG00000274890
......
Tcrlibrary_RUNX2_3_geneTcrlibrary_RUNX2_3_gene
Tcrlibrary_ZAP70_1_geneTcrlibrary_ZAP70_1_gene
Tcrlibrary_ZAP70_2_geneTcrlibrary_ZAP70_2_gene
Tcrlibrary_ZAP70_3_geneTcrlibrary_ZAP70_3_gene
Cas9_blast_geneCas9_blast_gene
\n", "

64370 rows × 1 columns

\n", "
" ], "text/plain": [ " gene_ids\n", "DDX11L1 ENSG00000223972\n", "WASH7P ENSG00000227232\n", "MIR6859-2 ENSG00000278267\n", "MIR1302-10 ENSG00000243485\n", "MIR1302-11 ENSG00000274890\n", "... ...\n", "Tcrlibrary_RUNX2_3_gene Tcrlibrary_RUNX2_3_gene\n", "Tcrlibrary_ZAP70_1_gene Tcrlibrary_ZAP70_1_gene\n", "Tcrlibrary_ZAP70_2_gene Tcrlibrary_ZAP70_2_gene\n", "Tcrlibrary_ZAP70_3_gene Tcrlibrary_ZAP70_3_gene\n", "Cas9_blast_gene Cas9_blast_gene\n", "\n", "[64370 rows x 1 columns]" ] }, "execution_count": 5, "metadata": {}, "output_type": "execute_result" } ], "source": [ "adata.var" ] }, { "cell_type": "code", "execution_count": 6, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "5AUbkDZFP2IK", "outputId": "1ca27838-348a-4b2a-98df-ef3de27cc4c8" }, "outputs": [], "source": [ "# We can normalize the data, however, it is better to avoid\n", "# preprocessing the whole dataset before splitting in training and test\n", "# to avoid data leakage.\n", "# NOTE: Normalization can be done inside the cross-val loop\n", "# sc.pp.normalize_total(adata, target_sum=1e6)\n", "\n", "# Log-transform the data does not leak data as it does not estimate anything\n", "sc.pp.log1p(adata)" ] }, { "cell_type": "code", "execution_count": 7, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 406 }, "id": "ugELTxTtQx9p", "outputId": "a476ce35-cc52-4d73-ff0e-b6487a1d95b6" }, "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", "
AAACCTGAGAAACCAT-1
AAACCTGAGAAACCGC-1
AAACCTGAGAAACCTA-1
AAACCTGAGAAACGAG-1
AAACCTGAGAAACGCC-1
...
TTTGTCATCTTTACAC-2
TTTGTCATCTTTACGT-2
TTTGTCATCTTTAGGG-2
TTTGTCATCTTTAGTC-2
TTTGTCATCTTTCCTC-2
\n", "

1474560 rows × 0 columns

\n", "
" ], "text/plain": [ "Empty DataFrame\n", "Columns: []\n", "Index: [AAACCTGAGAAACCAT-1, AAACCTGAGAAACCGC-1, AAACCTGAGAAACCTA-1, AAACCTGAGAAACGAG-1, AAACCTGAGAAACGCC-1, AAACCTGAGAAAGTGG-1, AAACCTGAGAACAACT-1, AAACCTGAGAACAATC-1, AAACCTGAGAACTCGG-1, AAACCTGAGAACTGTA-1, AAACCTGAGAAGAAGC-1, AAACCTGAGAAGATTC-1, AAACCTGAGAAGCCCA-1, AAACCTGAGAAGGACA-1, AAACCTGAGAAGGCCT-1, AAACCTGAGAAGGGTA-1, AAACCTGAGAAGGTGA-1, AAACCTGAGAAGGTTT-1, AAACCTGAGAATAGGG-1, AAACCTGAGAATCTCC-1, AAACCTGAGAATGTGT-1, AAACCTGAGAATGTTG-1, AAACCTGAGAATTCCC-1, AAACCTGAGAATTGTG-1, AAACCTGAGACAAAGG-1, AAACCTGAGACAAGCC-1, AAACCTGAGACAATAC-1, AAACCTGAGACACGAC-1, AAACCTGAGACACTAA-1, AAACCTGAGACAGACC-1, AAACCTGAGACAGAGA-1, AAACCTGAGACAGGCT-1, AAACCTGAGACATAAC-1, AAACCTGAGACCACGA-1, AAACCTGAGACCCACC-1, AAACCTGAGACCGGAT-1, AAACCTGAGACCTAGG-1, AAACCTGAGACCTTTG-1, AAACCTGAGACGACGT-1, AAACCTGAGACGCAAC-1, AAACCTGAGACGCACA-1, AAACCTGAGACGCTTT-1, AAACCTGAGACTAAGT-1, AAACCTGAGACTACAA-1, AAACCTGAGACTAGAT-1, AAACCTGAGACTAGGC-1, AAACCTGAGACTCGGA-1, AAACCTGAGACTGGGT-1, AAACCTGAGACTGTAA-1, AAACCTGAGACTTGAA-1, AAACCTGAGACTTTCG-1, AAACCTGAGAGAACAG-1, AAACCTGAGAGACGAA-1, AAACCTGAGAGACTAT-1, AAACCTGAGAGACTTA-1, AAACCTGAGAGAGCTC-1, AAACCTGAGAGATGAG-1, AAACCTGAGAGCAATT-1, AAACCTGAGAGCCCAA-1, AAACCTGAGAGCCTAG-1, AAACCTGAGAGCTATA-1, AAACCTGAGAGCTGCA-1, AAACCTGAGAGCTGGT-1, AAACCTGAGAGCTTCT-1, AAACCTGAGAGGACGG-1, AAACCTGAGAGGGATA-1, AAACCTGAGAGGGCTT-1, AAACCTGAGAGGTACC-1, AAACCTGAGAGGTAGA-1, AAACCTGAGAGGTTAT-1, AAACCTGAGAGGTTGC-1, AAACCTGAGAGTAAGG-1, AAACCTGAGAGTAATC-1, AAACCTGAGAGTACAT-1, AAACCTGAGAGTACCG-1, AAACCTGAGAGTCGGT-1, AAACCTGAGAGTCTGG-1, AAACCTGAGAGTGACC-1, AAACCTGAGAGTGAGA-1, AAACCTGAGAGTTGGC-1, AAACCTGAGATACACA-1, AAACCTGAGATAGCAT-1, AAACCTGAGATAGGAG-1, AAACCTGAGATAGTCA-1, AAACCTGAGATATACG-1, AAACCTGAGATATGCA-1, AAACCTGAGATATGGT-1, AAACCTGAGATCACGG-1, AAACCTGAGATCCCAT-1, AAACCTGAGATCCCGC-1, AAACCTGAGATCCGAG-1, AAACCTGAGATCCTGT-1, AAACCTGAGATCGATA-1, AAACCTGAGATCGGGT-1, AAACCTGAGATCTGAA-1, AAACCTGAGATCTGCT-1, AAACCTGAGATGAGAG-1, AAACCTGAGATGCCAG-1, AAACCTGAGATGCCTT-1, AAACCTGAGATGCGAC-1, ...]\n", "\n", "[1474560 rows x 0 columns]" ] }, "execution_count": 7, "metadata": {}, "output_type": "execute_result" } ], "source": [ "adata.obs" ] }, { "cell_type": "code", "execution_count": 8, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "f5utYHPCE_aT", "outputId": "1a32a812-9c3d-4bdc-ca0f-652943aad40e" }, "outputs": [ { "data": { "text/plain": [ "Index(['AAACCTGAGAAACCAT-1', 'AAACCTGAGAAACCGC-1', 'AAACCTGAGAAACCTA-1',\n", " 'AAACCTGAGAAACGAG-1', 'AAACCTGAGAAACGCC-1', 'AAACCTGAGAAAGTGG-1',\n", " 'AAACCTGAGAACAACT-1', 'AAACCTGAGAACAATC-1', 'AAACCTGAGAACTCGG-1',\n", " 'AAACCTGAGAACTGTA-1',\n", " ...\n", " 'TTTGTCATCTTGGGTA-2', 'TTTGTCATCTTGTACT-2', 'TTTGTCATCTTGTATC-2',\n", " 'TTTGTCATCTTGTCAT-2', 'TTTGTCATCTTGTTTG-2', 'TTTGTCATCTTTACAC-2',\n", " 'TTTGTCATCTTTACGT-2', 'TTTGTCATCTTTAGGG-2', 'TTTGTCATCTTTAGTC-2',\n", " 'TTTGTCATCTTTCCTC-2'],\n", " dtype='object', length=1474560)" ] }, "execution_count": 8, "metadata": {}, "output_type": "execute_result" } ], "source": [ "barcodes = adata.obs_names\n", "barcodes" ] }, { "cell_type": "code", "execution_count": 9, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "caIATAb6F6KO", "outputId": "46ae6909-b20d-438e-8cfc-24b26d43ffe2" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Index(['DDX11L1', 'WASH7P', 'MIR6859-2', 'MIR1302-10', 'MIR1302-11', 'FAM138A',\n", " 'OR4G4P', 'OR4G11P', 'OR4F5', 'RP11-34P13.7',\n", " ...\n", " 'Tcrlibrary_RUNX1_1_gene', 'Tcrlibrary_RUNX1_2_gene',\n", " 'Tcrlibrary_RUNX1_3_gene', 'Tcrlibrary_RUNX2_1_gene',\n", " 'Tcrlibrary_RUNX2_2_gene', 'Tcrlibrary_RUNX2_3_gene',\n", " 'Tcrlibrary_ZAP70_1_gene', 'Tcrlibrary_ZAP70_2_gene',\n", " 'Tcrlibrary_ZAP70_3_gene', 'Cas9_blast_gene'],\n", " dtype='object', length=64370)\n" ] } ], "source": [ "gene_names = adata.var.index\n", "print(gene_names)" ] }, { "cell_type": "code", "execution_count": 10, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "3_qlaVGLHLTU", "outputId": "937253d4-e12a-437c-ed3b-9077a4043fd8" }, "outputs": [ { "data": { "text/plain": [ "1735" ] }, "execution_count": 10, "metadata": {}, "output_type": "execute_result" } ], "source": [ "len(set(df_labels.barcode.tolist()))" ] }, { "cell_type": "code", "execution_count": 11, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "NudJw90wHM9B", "outputId": "8c46748f-d965-4706-f3d5-d5c8129fb2c0" }, "outputs": [ { "data": { "text/plain": [ "1474560" ] }, "execution_count": 11, "metadata": {}, "output_type": "execute_result" } ], "source": [ "len(set(barcodes.tolist()))" ] }, { "cell_type": "code", "execution_count": 12, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "-MUOcHm6HWS2", "outputId": "29d31d6c-9f11-46c2-e3da-8f2fde1c7f93" }, "outputs": [ { "data": { "text/plain": [ "1735" ] }, "execution_count": 12, "metadata": {}, "output_type": "execute_result" } ], "source": [ "matched_barcodes = sorted(set(barcodes.tolist()) & set(df_labels.barcode.tolist()))\n", "len(matched_barcodes)" ] }, { "cell_type": "code", "execution_count": 13, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 423 }, "id": "bH_1lBq6Lp8L", "outputId": "740058ff-2918-466b-9575-76c10db36701" }, "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", "
barcodeTCR
0AAACCTGCACACATGT-10
1AAACCTGCACGTCTCT-10
2AAACCTGTCAATACCG-10
3AAACCTGTCGTGGTCG-10
4AAACGGGTCTGAGTGT-10
.........
1730TTTCCTCGTCATGCCG-21
1731TTTGCGCGTAGCCTCG-21
1732TTTGGTTAGATACACA-21
1733TTTGGTTGTATGAATG-21
1734TTTGGTTTCCAAGTAC-21
\n", "

1735 rows × 2 columns

\n", "
" ], "text/plain": [ " barcode TCR\n", "0 AAACCTGCACACATGT-1 0\n", "1 AAACCTGCACGTCTCT-1 0\n", "2 AAACCTGTCAATACCG-1 0\n", "3 AAACCTGTCGTGGTCG-1 0\n", "4 AAACGGGTCTGAGTGT-1 0\n", "... ... ...\n", "1730 TTTCCTCGTCATGCCG-2 1\n", "1731 TTTGCGCGTAGCCTCG-2 1\n", "1732 TTTGGTTAGATACACA-2 1\n", "1733 TTTGGTTGTATGAATG-2 1\n", "1734 TTTGGTTTCCAAGTAC-2 1\n", "\n", "[1735 rows x 2 columns]" ] }, "execution_count": 13, "metadata": {}, "output_type": "execute_result" } ], "source": [ "#This is the InPathsY data in the original code of KPNNs\n", "df_labels" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Import PKN with CORNETO" ] }, { "cell_type": "code", "execution_count": 14, "metadata": {}, "outputs": [ { "data": { "text/html": [], "text/plain": [] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "\n", " \n", " \n", " \n", " \n", " \n", "
\n", " \n", " \n", " \n", " \n", "
Installed version:v1.0.0.dev1 (latest stable: v1.0.0-alpha)
Available backends:CVXPY v1.6.0
Default backend (corneto.opt):CVXPY
Installed solvers:CLARABEL, CVXOPT, GLPK, GLPK_MI, GUROBI, HIGHS, SCIP, SCIPY
Graphviz version:v0.20.3
Installed path:/Users/pablorodriguezmier/Documents/work/projects/corneto/corneto
Repository:https://github.com/saezlab/corneto
\n", "
" ], "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "import corneto as cn\n", "cn.info()" ] }, { "cell_type": "code", "execution_count": 15, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(13121, 1)" ] }, "execution_count": 15, "metadata": {}, "output_type": "execute_result" } ], "source": [ "outputs_pkn = list(set(df_edges.parent.tolist()) - set(df_edges.child.tolist()))\n", "inputs_pkn = set(df_edges.child.tolist()) - set(df_edges.parent.tolist())\n", "input_pkn_genes = list(set(g.split(\"_\")[0] for g in inputs_pkn))\n", "len(inputs_pkn), len(outputs_pkn)" ] }, { "cell_type": "code", "execution_count": 16, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(13439, 27579)" ] }, "execution_count": 16, "metadata": {}, "output_type": "execute_result" } ], "source": [ "tuples = [(r.child, 1, r.parent) for _, r in df_edges.iterrows()]\n", "G = cn.Graph.from_sif_tuples(tuples)\n", "G = G.prune(inputs_pkn, outputs_pkn)\n", "\n", "# Size of the original PKN provided by the authors\n", "G.shape" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Select the single cell data for training" ] }, { "cell_type": "code", "execution_count": 17, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "XCy2cd5zK2cp", "outputId": "5307650d-0633-48fd-a086-d2a64ef1b1e7" }, "outputs": [ { "data": { "text/plain": [ "(1735, 14229)" ] }, "execution_count": 17, "metadata": {}, "output_type": "execute_result" } ], "source": [ "adata_matched = adata[adata.obs_names.isin(matched_barcodes), adata.var_names.isin(input_pkn_genes)]\n", "adata_matched.shape" ] }, { "cell_type": "code", "execution_count": 18, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "L0qiukuPK-ky", "outputId": "91d4bb75-f51c-46be-e270-22afd8d73d36" }, "outputs": [ { "data": { "text/plain": [ "12459" ] }, "execution_count": 18, "metadata": {}, "output_type": "execute_result" } ], "source": [ "non_zero_genes = set(adata_matched.to_df().columns[adata_matched.to_df().sum(axis=0) >= 1e-6].values)\n", "len(non_zero_genes)" ] }, { "cell_type": "code", "execution_count": 19, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "l2Ia3ViolksN", "outputId": "e8a9e21c-c52a-48f6-a213-89b88f2df238" }, "outputs": [ { "data": { "text/plain": [ "12459" ] }, "execution_count": 19, "metadata": {}, "output_type": "execute_result" } ], "source": [ "len(non_zero_genes.intersection(adata_matched.var_names))" ] }, { "cell_type": "code", "execution_count": 20, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "dKPAE_g4hefN", "outputId": "c1a5b3e1-c627-4ac5-e390-19061fb32cfc" }, "outputs": [ { "data": { "text/plain": [ "(1735, 12487)" ] }, "execution_count": 20, "metadata": {}, "output_type": "execute_result" } ], "source": [ "adata_matched = adata_matched[:, adata_matched.var_names.isin(non_zero_genes)]\n", "# Many duplicates still 0 counts\n", "adata_matched = adata_matched[:, adata_matched.to_df().sum(axis=0) != 0]\n", "adata_matched.shape" ] }, { "cell_type": "code", "execution_count": 21, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 617 }, "id": "-dVxay1DjH0X", "outputId": "7a0816fc-cbee-48eb-ecdb-62a0e9d18488" }, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "/var/folders/b4/gwkwsdb93sv11rtztqbm3l040000gn/T/ipykernel_91875/3016726407.py:2: FutureWarning: DataFrame.groupby with axis=1 is deprecated. Do `frame.T.groupby(...)` without axis instead.\n", " df_expr = df_expr.groupby(df_expr.columns, axis=1).max()\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", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
A1BGA2ML1AAASAACSAADATAAED1AAGABAAK1AAMDCAAMP...ZSWIM8ZUFSPZW10ZWILCHZXDCZYG11AZYG11BZYXZZEF1ZZZ3
AAACCTGCACACATGT-10.00.00.0000000.6931470.00.0000000.0000000.0000000.01.098612...0.0000000.0000000.0000000.0000000.00.00.00.0000000.0000000.000000
AAACCTGCACGTCTCT-10.00.00.0000000.0000000.00.6931470.0000000.0000000.00.693147...0.0000000.0000000.0000000.0000000.00.00.00.0000000.0000000.000000
AAACCTGTCAATACCG-10.00.00.0000000.0000000.00.0000000.0000000.0000000.00.693147...0.0000000.0000000.0000000.0000000.00.00.00.0000000.0000000.000000
AAACCTGTCGTGGTCG-10.00.01.0986120.0000000.00.0000000.0000000.0000000.01.098612...0.0000000.6931470.0000000.6931470.00.00.00.0000000.0000000.000000
AAACGGGTCTGAGTGT-10.00.00.0000000.0000000.00.6931470.0000000.0000000.00.000000...0.6931470.0000000.6931470.0000000.00.00.00.0000000.0000000.000000
..................................................................
TTTCCTCGTCATGCCG-20.00.00.0000000.0000000.00.0000000.0000000.6931470.00.693147...0.0000000.0000000.0000000.0000000.00.00.00.0000000.0000000.000000
TTTGCGCGTAGCCTCG-20.00.01.0986120.0000000.00.0000000.0000000.0000000.01.386294...0.0000000.0000000.0000000.0000000.00.00.00.0000000.0000000.000000
TTTGGTTAGATACACA-20.00.01.0986120.0000000.00.0000000.0000001.0986120.00.693147...0.6931470.0000000.6931470.0000000.00.00.00.6931470.6931470.000000
TTTGGTTGTATGAATG-20.00.00.0000000.0000000.00.6931471.0986120.0000000.00.693147...0.0000000.6931470.0000000.0000000.00.00.00.0000000.0000000.693147
TTTGGTTTCCAAGTAC-20.00.00.0000000.0000000.00.0000000.0000000.0000000.00.693147...0.0000000.0000000.0000000.0000000.00.00.01.0986120.0000000.000000
\n", "

1735 rows × 12459 columns

\n", "
" ], "text/plain": [ " A1BG A2ML1 AAAS AACS AADAT AAED1 \\\n", "AAACCTGCACACATGT-1 0.0 0.0 0.000000 0.693147 0.0 0.000000 \n", "AAACCTGCACGTCTCT-1 0.0 0.0 0.000000 0.000000 0.0 0.693147 \n", "AAACCTGTCAATACCG-1 0.0 0.0 0.000000 0.000000 0.0 0.000000 \n", "AAACCTGTCGTGGTCG-1 0.0 0.0 1.098612 0.000000 0.0 0.000000 \n", "AAACGGGTCTGAGTGT-1 0.0 0.0 0.000000 0.000000 0.0 0.693147 \n", "... ... ... ... ... ... ... \n", "TTTCCTCGTCATGCCG-2 0.0 0.0 0.000000 0.000000 0.0 0.000000 \n", "TTTGCGCGTAGCCTCG-2 0.0 0.0 1.098612 0.000000 0.0 0.000000 \n", "TTTGGTTAGATACACA-2 0.0 0.0 1.098612 0.000000 0.0 0.000000 \n", "TTTGGTTGTATGAATG-2 0.0 0.0 0.000000 0.000000 0.0 0.693147 \n", "TTTGGTTTCCAAGTAC-2 0.0 0.0 0.000000 0.000000 0.0 0.000000 \n", "\n", " AAGAB AAK1 AAMDC AAMP ... ZSWIM8 \\\n", "AAACCTGCACACATGT-1 0.000000 0.000000 0.0 1.098612 ... 0.000000 \n", "AAACCTGCACGTCTCT-1 0.000000 0.000000 0.0 0.693147 ... 0.000000 \n", "AAACCTGTCAATACCG-1 0.000000 0.000000 0.0 0.693147 ... 0.000000 \n", "AAACCTGTCGTGGTCG-1 0.000000 0.000000 0.0 1.098612 ... 0.000000 \n", "AAACGGGTCTGAGTGT-1 0.000000 0.000000 0.0 0.000000 ... 0.693147 \n", "... ... ... ... ... ... ... \n", "TTTCCTCGTCATGCCG-2 0.000000 0.693147 0.0 0.693147 ... 0.000000 \n", "TTTGCGCGTAGCCTCG-2 0.000000 0.000000 0.0 1.386294 ... 0.000000 \n", "TTTGGTTAGATACACA-2 0.000000 1.098612 0.0 0.693147 ... 0.693147 \n", "TTTGGTTGTATGAATG-2 1.098612 0.000000 0.0 0.693147 ... 0.000000 \n", "TTTGGTTTCCAAGTAC-2 0.000000 0.000000 0.0 0.693147 ... 0.000000 \n", "\n", " ZUFSP ZW10 ZWILCH ZXDC ZYG11A ZYG11B \\\n", "AAACCTGCACACATGT-1 0.000000 0.000000 0.000000 0.0 0.0 0.0 \n", "AAACCTGCACGTCTCT-1 0.000000 0.000000 0.000000 0.0 0.0 0.0 \n", "AAACCTGTCAATACCG-1 0.000000 0.000000 0.000000 0.0 0.0 0.0 \n", "AAACCTGTCGTGGTCG-1 0.693147 0.000000 0.693147 0.0 0.0 0.0 \n", "AAACGGGTCTGAGTGT-1 0.000000 0.693147 0.000000 0.0 0.0 0.0 \n", "... ... ... ... ... ... ... \n", "TTTCCTCGTCATGCCG-2 0.000000 0.000000 0.000000 0.0 0.0 0.0 \n", "TTTGCGCGTAGCCTCG-2 0.000000 0.000000 0.000000 0.0 0.0 0.0 \n", "TTTGGTTAGATACACA-2 0.000000 0.693147 0.000000 0.0 0.0 0.0 \n", "TTTGGTTGTATGAATG-2 0.693147 0.000000 0.000000 0.0 0.0 0.0 \n", "TTTGGTTTCCAAGTAC-2 0.000000 0.000000 0.000000 0.0 0.0 0.0 \n", "\n", " ZYX ZZEF1 ZZZ3 \n", "AAACCTGCACACATGT-1 0.000000 0.000000 0.000000 \n", "AAACCTGCACGTCTCT-1 0.000000 0.000000 0.000000 \n", "AAACCTGTCAATACCG-1 0.000000 0.000000 0.000000 \n", "AAACCTGTCGTGGTCG-1 0.000000 0.000000 0.000000 \n", "AAACGGGTCTGAGTGT-1 0.000000 0.000000 0.000000 \n", "... ... ... ... \n", "TTTCCTCGTCATGCCG-2 0.000000 0.000000 0.000000 \n", "TTTGCGCGTAGCCTCG-2 0.000000 0.000000 0.000000 \n", "TTTGGTTAGATACACA-2 0.693147 0.693147 0.000000 \n", "TTTGGTTGTATGAATG-2 0.000000 0.000000 0.693147 \n", "TTTGGTTTCCAAGTAC-2 1.098612 0.000000 0.000000 \n", "\n", "[1735 rows x 12459 columns]" ] }, "execution_count": 21, "metadata": {}, "output_type": "execute_result" } ], "source": [ "df_expr = adata_matched.to_df()\n", "df_expr = df_expr.groupby(df_expr.columns, axis=1).max()\n", "df_expr" ] }, { "cell_type": "markdown", "metadata": { "id": "RrIZBw2PqKQ2" }, "source": [ "## Building and training the KPNN\n", "\n", "Now we will use the provided PKN by the authors and the utility functions in CORNETO to build a KPNN similar to the one used in the original manuscript" ] }, { "cell_type": "code", "execution_count": 22, "metadata": {}, "outputs": [], "source": [ "import os\n", "os.environ[\"KERAS_BACKEND\"] = \"jax\"" ] }, { "cell_type": "code", "execution_count": 23, "metadata": { "id": "5X0GxeLeqLRK" }, "outputs": [], "source": [ "import keras\n", "from sklearn.model_selection import StratifiedKFold\n", "from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score, roc_auc_score\n", "from keras.optimizers import Adam\n", "from keras.callbacks import EarlyStopping\n", "import numpy as np" ] }, { "cell_type": "code", "execution_count": 24, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "I9PoXtGOtvYs", "outputId": "c22ed990-e595-4b6e-e4c4-da4ba061b93a" }, "outputs": [ { "data": { "text/plain": [ "((1735, 12459), (1735,))" ] }, "execution_count": 24, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Use the data from the experiment\n", "X = df_expr.values\n", "y = df_labels.set_index(\"barcode\").loc[df_expr.index, \"TCR\"].values\n", "X.shape, y.shape" ] }, { "cell_type": "code", "execution_count": 25, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(13121, 1)" ] }, "execution_count": 25, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# We can prefilter on top N genes to make this faster\n", "top_n = None\n", "\n", "# From the given PKN\n", "outputs_pkn = list(set(df_edges.parent.tolist()) - set(df_edges.child.tolist()))\n", "inputs_pkn = set(df_edges.child.tolist()) - set(df_edges.parent.tolist())\n", "input_pkn_genes = list(set(g.split(\"_\")[0] for g in inputs_pkn))\n", "\n", "if top_n is not None and top_n > 0:\n", " input_pkn_genes = list(set(input_pkn_genes).intersection(df_expr.var(axis=0).sort_values(ascending=False).head(top_n).index))\n", " inputs_pkn = list(g + \"_gene\" for g in input_pkn_genes)\n", "\n", "len(inputs_pkn), len(outputs_pkn)" ] }, { "cell_type": "code", "execution_count": 26, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "12459" ] }, "execution_count": 26, "metadata": {}, "output_type": "execute_result" } ], "source": [ "input_nn_genes = list(set(input_pkn_genes).intersection(df_expr.columns))\n", "input_nn = [g + \"_gene\" for g in input_nn_genes]\n", "len(input_nn)" ] }, { "cell_type": "code", "execution_count": 27, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(12767, 25928)" ] }, "execution_count": 27, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Build corneto graph\n", "tuples = [(r.child, 1, r.parent) for _, r in df_edges.iterrows()]\n", "G = cn.Graph.from_sif_tuples(tuples)\n", "G = G.prune(input_nn, outputs_pkn)\n", "G.shape" ] }, { "cell_type": "code", "execution_count": 28, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(12459, 12459)" ] }, "execution_count": 28, "metadata": {}, "output_type": "execute_result" } ], "source": [ "len(input_nn), len(input_nn_genes)" ] }, { "cell_type": "code", "execution_count": 29, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "12459" ] }, "execution_count": 29, "metadata": {}, "output_type": "execute_result" } ], "source": [ "len(set(input_nn).intersection(G.V))" ] }, { "cell_type": "code", "execution_count": 30, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "((1735, 12459), (1735,))" ] }, "execution_count": 30, "metadata": {}, "output_type": "execute_result" } ], "source": [ "X = df_expr.loc[:, input_nn_genes].values\n", "y = df_labels.set_index(\"barcode\").loc[df_expr.index, \"TCR\"].values\n", "X.shape, y.shape" ] }, { "cell_type": "code", "execution_count": 31, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Building DAG NN model with CORNETO using Keras with JAX...\n", " > N. inputs: 12459\n", " > N. outputs: 1\n", " > N. parameters: 26236\n", "Compiling...\n", "Fitting...\n", "Weights saved to /var/folders/b4/gwkwsdb93sv11rtztqbm3l040000gn/T/tmpj5f_9545/weights_0.keras\n", "\u001b[1m11/11\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 208ms/step\n", " > Fold 0 validation ROC-AUC=0.991\n", "Building DAG NN model with CORNETO using Keras with JAX...\n", " > N. inputs: 12459\n", " > N. outputs: 1\n", " > N. parameters: 26236\n", "Compiling...\n", "Fitting...\n", "Weights saved to /var/folders/b4/gwkwsdb93sv11rtztqbm3l040000gn/T/tmpj5f_9545/weights_1.keras\n", "\u001b[1m11/11\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 180ms/step\n", " > Fold 1 validation ROC-AUC=0.983\n", "Building DAG NN model with CORNETO using Keras with JAX...\n", " > N. inputs: 12459\n", " > N. outputs: 1\n", " > N. parameters: 26236\n", "Compiling...\n", "Fitting...\n", "Weights saved to /var/folders/b4/gwkwsdb93sv11rtztqbm3l040000gn/T/tmpj5f_9545/weights_2.keras\n", "\u001b[1m11/11\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 179ms/step\n", " > Fold 2 validation ROC-AUC=0.986\n", "Building DAG NN model with CORNETO using Keras with JAX...\n", " > N. inputs: 12459\n", " > N. outputs: 1\n", " > N. parameters: 26236\n", "Compiling...\n", "Fitting...\n", "Weights saved to /var/folders/b4/gwkwsdb93sv11rtztqbm3l040000gn/T/tmpj5f_9545/weights_3.keras\n", "\u001b[1m11/11\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 180ms/step\n", " > Fold 3 validation ROC-AUC=0.993\n", "Building DAG NN model with CORNETO using Keras with JAX...\n", " > N. inputs: 12459\n", " > N. outputs: 1\n", " > N. parameters: 26236\n", "Compiling...\n", "Fitting...\n", "Weights saved to /var/folders/b4/gwkwsdb93sv11rtztqbm3l040000gn/T/tmpj5f_9545/weights_4.keras\n", "\u001b[1m11/11\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 181ms/step\n", " > Fold 4 validation ROC-AUC=0.994\n", "Validation metrics:\n", " - accuracy: 0.957\n", " - precision: 0.961\n", " - recall: 0.951\n", " - f1: 0.956\n", " - roc_auc: 0.989\n" ] } ], "source": [ "from corneto._ml import build_dagnn\n", "\n", "def stratified_kfold(\n", " G,\n", " inputs,\n", " outputs,\n", " n_splits=5,\n", " shuffle=True,\n", " random_state=42,\n", " lr=0.001,\n", " patience=10,\n", " file_weights=\"weights\",\n", " dagnn_config=dict(\n", " batch_norm_input=True,\n", " batch_norm_center=False,\n", " batch_norm_scale=False,\n", " bias_reg_l1=1e-3,\n", " bias_reg_l2=1e-2,\n", " dropout=0.20,\n", " default_hidden_activation=\"sigmoid\",\n", " default_output_activation=\"sigmoid\",\n", " verbose=False\n", " )\n", "):\n", " kfold = StratifiedKFold(n_splits=n_splits, shuffle=shuffle, random_state=random_state)\n", " models = []\n", " metrics = {m: [] for m in [\"accuracy\", \"precision\", \"recall\", \"f1\", \"roc_auc\"]}\n", " for i, (train_idx, val_idx) in enumerate(kfold.split(X, y)):\n", " X_train, X_val = X[train_idx], X[val_idx]\n", " y_train, y_val = y[train_idx], y[val_idx]\n", " \n", " print(\"Building DAG NN model with CORNETO using Keras with JAX...\")\n", " print(f\" > N. inputs: {len(input_nn)}\")\n", " print(f\" > N. outputs: {len(outputs_pkn)}\")\n", " model = build_dagnn(\n", " G, \n", " input_nn, \n", " outputs_pkn,\n", " **dagnn_config\n", " )\n", " print(f\" > N. parameters: {model.count_params()}\")\n", " \n", " # Train the model with Adam\n", " opt=keras.optimizers.Adam(learning_rate=lr)\n", " early_stopping = EarlyStopping(monitor='val_loss', patience=patience, restore_best_weights=True)\n", " print(\"Compiling...\")\n", " model.compile(\n", " optimizer=opt,\n", " loss='binary_crossentropy',\n", " metrics=['accuracy']\n", " )\n", " print(\"Fitting...\")\n", " model.fit(X_train, y_train,\n", " validation_data=(X_val, y_val),\n", " epochs=200,\n", " batch_size=64,\n", " verbose=0,\n", " callbacks=[early_stopping])\n", " \n", " if file_weights is not None:\n", " filename = f\"{file_weights}_{i}.keras\"\n", " model.save(filename)\n", " print(f\"Weights saved to {filename}\")\n", " \n", " # Predictions and metrics calculation\n", " y_pred_proba = model.predict(X_val).flatten()\n", " y_pred = (y_pred_proba > 0.5).astype(int)\n", " acc = accuracy_score(y_val, y_pred)\n", " precision = precision_score(y_val, y_pred)\n", " recall = recall_score(y_val, y_pred)\n", " f1 = f1_score(y_val, y_pred)\n", " roc_auc = roc_auc_score(y_val, y_pred_proba)\n", " metrics[\"accuracy\"].append(acc)\n", " metrics[\"precision\"].append(precision)\n", " metrics[\"recall\"].append(recall)\n", " metrics[\"f1\"].append(f1)\n", " metrics[\"roc_auc\"].append(roc_auc)\n", " print(f\" > Fold {i} validation ROC-AUC={roc_auc:.3f}\")\n", " models.append(model)\n", " return models, metrics\n", "\n", "temp_weights = tempfile.mkdtemp()\n", "models, metrics = stratified_kfold(G, input_nn, outputs_pkn, file_weights=os.path.join(temp_weights, \"weights\"))\n", "\n", "print(\"Validation metrics:\")\n", "for k, v in metrics.items():\n", " print(f\" - {k}: {np.mean(v):.3f}\")" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Now, we will analyze the learned biases for each of the nodes in the graph. Note that authors in the KPNN paper explain a way to extract weights for the nodes, based on the learned interactions and accounting for biases in the structure of the NN. Here we just show the learned biases of the nodes of the NN across 5 folds. Please be careful interpreting these weights." ] }, { "cell_type": "code", "execution_count": 32, "metadata": {}, "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", "
biasabs_biaspow2_bias
foldgene
1DUSP3-2.6327342.6327346.931287
2PRKCD-2.1976722.1976724.829763
SUZ12.EZH2-2.1261502.1261504.520514
NfKb.p65.p50-2.0106822.0106824.042840
TCR1.9883341.9883343.953472
4ZAP70-1.9522611.9522613.811324
3YAP1-1.8720041.8720043.504399
4NfKb.p65.p50-1.8347471.8347473.366298
0PRC2-1.8266271.8266273.336568
4TCR1.8165161.8165163.299730
\n", "
" ], "text/plain": [ " bias abs_bias pow2_bias\n", "fold gene \n", "1 DUSP3 -2.632734 2.632734 6.931287\n", "2 PRKCD -2.197672 2.197672 4.829763\n", " SUZ12.EZH2 -2.126150 2.126150 4.520514\n", " NfKb.p65.p50 -2.010682 2.010682 4.042840\n", " TCR 1.988334 1.988334 3.953472\n", "4 ZAP70 -1.952261 1.952261 3.811324\n", "3 YAP1 -1.872004 1.872004 3.504399\n", "4 NfKb.p65.p50 -1.834747 1.834747 3.366298\n", "0 PRC2 -1.826627 1.826627 3.336568\n", "4 TCR 1.816516 1.816516 3.299730" ] }, "execution_count": 32, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# We collect the weights obtained in each fold\n", "\n", "def load_biases(file=\"weights\", folds=5):\n", " biases = []\n", " mean_inputs = []\n", " for i in range(5):\n", " model = keras.models.load_model(f\"{file}_{i}.keras\")\n", " for layer in model.layers:\n", " weights = layer.get_weights()\n", " if weights:\n", " biases.append((i, layer.name, weights[1][0]))\n", " mean_inputs.append((i, layer.name, weights[0].mean()))\n", " df_biases = pd.DataFrame(biases, columns=[\"fold\", \"gene\", \"bias\"])\n", " df_biases[\"abs_bias\"] = df_biases.bias.abs()\n", " df_biases[\"pow2_bias\"] = df_biases.bias.pow(2)\n", " df_biases = df_biases.set_index([\"fold\",\"gene\"])\n", " return df_biases\n", "\n", "df_biases = load_biases(file=os.path.join(temp_weights, \"weights\"), folds=5)\n", "df_biases.sort_values(by=\"abs_bias\", ascending=False).head(10)" ] }, { "cell_type": "code", "execution_count": 33, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "gene\n", "TCR 3.129825\n", "NfKb.p65.p50 2.217130\n", "PRKCD 2.162571\n", "PRC2 2.140556\n", "DUSP3 1.901882\n", " ... \n", "MLL2.complex 0.000641\n", "HSF2 0.000268\n", "GATA3 0.000145\n", "SRY 0.000138\n", "REST 0.000032\n", "Name: pow2_bias, Length: 308, dtype: float32" ] }, "execution_count": 33, "metadata": {}, "output_type": "execute_result" } ], "source": [ "df_biases_full = df_biases.copy().reset_index()\n", "gene_biases_score = df_biases_full.groupby(\"gene\")[\"pow2_bias\"].mean().sort_values(ascending=False)\n", "gene_biases_score" ] }, { "cell_type": "code", "execution_count": 34, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "" ] }, "execution_count": 34, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "gene_biases_score.head(30).plot.bar()" ] }, { "cell_type": "code", "execution_count": 35, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "import pandas as pd\n", "import seaborn as sns\n", "import matplotlib.pyplot as plt\n", "\n", "# Get the top genes sorted by mean bias\n", "top_genes = gene_biases_score.head(30).index\n", "\n", "# Filter the DataFrame to include only rows with these top genes\n", "filtered_df = df_biases_full[df_biases_full['gene'].isin(top_genes)]\n", "\n", "plt.figure(figsize=(10, 6))\n", "sns.violinplot(data=filtered_df, x=\"gene\", y=\"bias\", order=top_genes, dodge=True)\n", "sns.stripplot(data=filtered_df, x=\"gene\", y=\"bias\", order=top_genes, dodge=True)\n", "plt.xticks(rotation=90)\n", "plt.title('Biases of the trained neurons')\n", "plt.xlabel('Gene')\n", "plt.ylabel('Bias')\n", "plt.axhline(0, linestyle=\"--\", color=\"k\")\n", "plt.tight_layout()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Use CORNETO for NN pruning\n", "\n", "Now we will show how CORNETO can be used to extract a smaller, yet complete DAG from the original PKN provided by the authors. We will add input edges to each input node and an output edge through TCR to indicate which nodes are the inputs and which one the output. We will use then Acyclic Flow to find the smallest DAG comprising these nodes" ] }, { "cell_type": "code", "execution_count": 36, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "(12767, 38388)\n", "===============================================================================\n", " CVXPY \n", " v1.6.0 \n", "===============================================================================\n", "(CVXPY) Dec 23 01:19:36 PM: Your problem has 127931 variables, 332945 constraints, and 0 parameters.\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "/Users/pablorodriguezmier/miniforge3/envs/corneto/lib/python3.12/site-packages/cvxpy/problems/problem.py:158: UserWarning: Objective contains too many subexpressions. Consider vectorizing your CVXPY code to speed up compilation.\n", " warnings.warn(\"Objective contains too many subexpressions. \"\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "(CVXPY) Dec 23 01:19:36 PM: It is compliant with the following grammars: DCP, DQCP\n", "(CVXPY) Dec 23 01:19:36 PM: (If you need to solve this problem multiple times, but with different data, consider using parameters.)\n", "(CVXPY) Dec 23 01:19:36 PM: CVXPY will first compile your problem; then, it will invoke a numerical solver to obtain a solution.\n", "(CVXPY) Dec 23 01:19:36 PM: Your problem is compiled with the CPP canonicalization backend.\n", "-------------------------------------------------------------------------------\n", " Compilation \n", "-------------------------------------------------------------------------------\n", "(CVXPY) Dec 23 01:19:37 PM: Compiling problem (target solver=SCIP).\n", "(CVXPY) Dec 23 01:19:37 PM: Reduction chain: Dcp2Cone -> CvxAttr2Constr -> ConeMatrixStuffing -> SCIP\n", "(CVXPY) Dec 23 01:19:37 PM: Applying reduction Dcp2Cone\n", "(CVXPY) Dec 23 01:19:37 PM: Applying reduction CvxAttr2Constr\n", "(CVXPY) Dec 23 01:19:37 PM: Applying reduction ConeMatrixStuffing\n", "(CVXPY) Dec 23 01:20:56 PM: Applying reduction SCIP\n", "(CVXPY) Dec 23 01:20:56 PM: Finished problem compilation (took 7.962e+01 seconds).\n", "-------------------------------------------------------------------------------\n", " Numerical solver \n", "-------------------------------------------------------------------------------\n", "(CVXPY) Dec 23 01:20:56 PM: Invoking solver SCIP to obtain a solution.\n", "presolving:\n", "(round 1, fast) 72218 del vars, 263764 del conss, 0 add conss, 158463 chg bounds, 17010 chg sides, 17010 chg coeffs, 0 upgd conss, 0 impls, 0 clqs\n", "(round 2, fast) 83111 del vars, 288686 del conss, 0 add conss, 158636 chg bounds, 17982 chg sides, 17982 chg coeffs, 0 upgd conss, 0 impls, 0 clqs\n", "(round 3, fast) 83299 del vars, 288858 del conss, 0 add conss, 158667 chg bounds, 18157 chg sides, 18150 chg coeffs, 0 upgd conss, 0 impls, 0 clqs\n", "(round 4, fast) 83323 del vars, 288959 del conss, 0 add conss, 158670 chg bounds, 18308 chg sides, 18242 chg coeffs, 0 upgd conss, 0 impls, 0 clqs\n", "(round 5, fast) 87954 del vars, 288998 del conss, 0 add conss, 158670 chg bounds, 18310 chg sides, 18242 chg coeffs, 0 upgd conss, 0 impls, 0 clqs\n", "(round 6, fast) 87954 del vars, 288998 del conss, 0 add conss, 169675 chg bounds, 18310 chg sides, 18242 chg coeffs, 0 upgd conss, 0 impls, 0 clqs\n", "(round 7, fast) 87976 del vars, 289373 del conss, 0 add conss, 169675 chg bounds, 19713 chg sides, 29203 chg coeffs, 0 upgd conss, 0 impls, 0 clqs\n", "(round 8, fast) 87993 del vars, 289470 del conss, 0 add conss, 169675 chg bounds, 19713 chg sides, 29203 chg coeffs, 0 upgd conss, 0 impls, 0 clqs\n", "(round 9, exhaustive) 87994 del vars, 300376 del conss, 0 add conss, 169708 chg bounds, 19713 chg sides, 29203 chg coeffs, 0 upgd conss, 0 impls, 0 clqs\n", "(round 10, fast) 98924 del vars, 311359 del conss, 0 add conss, 169708 chg bounds, 19722 chg sides, 29210 chg coeffs, 0 upgd conss, 4004 impls, 0 clqs\n", "(round 11, fast) 99966 del vars, 312534 del conss, 0 add conss, 169708 chg bounds, 19722 chg sides, 29210 chg coeffs, 0 upgd conss, 4004 impls, 0 clqs\n", "(round 12, fast) 99978 del vars, 312581 del conss, 0 add conss, 169709 chg bounds, 19722 chg sides, 29210 chg coeffs, 0 upgd conss, 4004 impls, 0 clqs\n", "(round 13, exhaustive) 99980 del vars, 312589 del conss, 0 add conss, 169715 chg bounds, 19724 chg sides, 29212 chg coeffs, 9689 upgd conss, 4004 impls, 0 clqs\n", "(round 14, fast) 99988 del vars, 312611 del conss, 0 add conss, 169715 chg bounds, 19724 chg sides, 29212 chg coeffs, 9689 upgd conss, 14702 impls, 925 clqs\n", "(round 15, exhaustive) 99999 del vars, 312621 del conss, 0 add conss, 169718 chg bounds, 19726 chg sides, 29214 chg coeffs, 17411 upgd conss, 14704 impls, 908 clqs\n", "(round 16, medium) 100057 del vars, 312642 del conss, 0 add conss, 169718 chg bounds, 19726 chg sides, 29214 chg coeffs, 17411 upgd conss, 22425 impls, 1333 clqs\n", "(round 17, medium) 100059 del vars, 312756 del conss, 0 add conss, 169718 chg bounds, 19726 chg sides, 29214 chg coeffs, 17411 upgd conss, 22425 impls, 1333 clqs\n", "(round 18, exhaustive) 100754 del vars, 312760 del conss, 0 add conss, 169720 chg bounds, 19726 chg sides, 29214 chg coeffs, 17413 upgd conss, 22425 impls, 1333 clqs\n", "(round 19, fast) 100756 del vars, 312828 del conss, 0 add conss, 169720 chg bounds, 19726 chg sides, 29219 chg coeffs, 17413 upgd conss, 22427 impls, 1337 clqs\n", " (3.0s) probing: 1000/17563 (5.7%) - 0 fixings, 1 aggregations, 370 implications, 2 bound changes\n", " (3.0s) probing: 1003/17563 (5.7%) - 0 fixings, 1 aggregations, 370 implications, 2 bound changes\n", " (3.0s) probing aborted: 1000/1000 successive useless probings\n", " (3.0s) symmetry computation started: requiring (bin +, int +, cont +), (fixed: bin -, int -, cont -)\n", " (26.3s) symmetry computation finished: 1500 generators found (max: 1500, log10 of symmetry group size: 1750.0) (symcode time: 23.21)\n", "dynamic symmetry handling statistics:\n", " orbitopal reduction: 5 components: 3x3, 3x3, 3x3, 3x3, 4x3\n", " orbital reduction: 11 components of sizes 3, 5, 9, 4, 4, 4, 18, 17, 8, 6, 12\n", " lexicographic reduction: 105 permutations with support sizes 8, 6, 8, 6, 6, 8, 6, 6, 6, 8, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 8, 8, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 8, 8, 8, 8, 8, 8, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6\n", "handled 45 out of 45 symmetry components\n", "(round 20, exhaustive) 100757 del vars, 312828 del conss, 1385 add conss, 169722 chg bounds, 19726 chg sides, 29219 chg coeffs, 17417 upgd conss, 22520 impls, 1612 clqs\n", "(round 21, exhaustive) 101738 del vars, 312830 del conss, 1385 add conss, 169722 chg bounds, 19726 chg sides, 29219 chg coeffs, 17417 upgd conss, 22528 impls, 1616 clqs\n", "(round 22, fast) 101738 del vars, 313811 del conss, 1385 add conss, 169722 chg bounds, 19726 chg sides, 29219 chg coeffs, 17417 upgd conss, 22528 impls, 1616 clqs\n", " (26.5s) probing: 1103/17563 (6.3%) - 0 fixings, 1 aggregations, 563 implications, 2 bound changes\n", " (26.5s) probing aborted: 1000/1000 successive useless probings\n", "presolving (23 rounds: 23 fast, 9 medium, 7 exhaustive):\n", " 101738 deleted vars, 313811 deleted constraints, 1385 added constraints, 169722 tightened bounds, 0 added holes, 19726 changed sides, 29219 changed coefficients\n", " 22621 implications, 1716 cliques\n", "presolved problem has 26193 variables (17562 bin, 0 int, 0 impl, 8631 cont) and 20519 constraints\n", " 16888 constraints of type \n", " 3323 constraints of type \n", " 308 constraints of type \n", "transformed objective value is always integral (scale: 1)\n", "Presolving Time: 26.29\n", "\n", " time | node | left |LP iter|LP it/n|mem/heur|mdpt |vars |cons |rows |cuts |sepa|confs|strbr| dualbound | primalbound | gap | compl. \n", "p26.7s| 1 | 0 | 115 | - | locks| 0 | 26k| 20k| 20k| 0 | 0 | 0 | 0 | 2.066500e+04 | 2.744200e+04 | 32.79%| unknown\n", "i26.8s| 1 | 0 | 115 | - | oneopt| 0 | 26k| 20k| 20k| 0 | 0 | 24 | 0 | 2.066500e+04 | 2.707300e+04 | 31.01%| unknown\n", " 27.6s| 1 | 0 | 11865 | - | 661M | 0 | 26k| 20k| 20k| 0 | 0 | 24 | 0 | 2.097314e+04 | 2.707300e+04 | 29.08%| unknown\n", " 28.4s| 1 | 0 | 16422 | - | 671M | 0 | 26k| 20k| 21k|1624 | 1 | 24 | 0 | 2.258309e+04 | 2.707300e+04 | 19.88%| unknown\n", " 29.4s| 1 | 0 | 21208 | - | 674M | 0 | 26k| 20k| 21k|1857 | 2 | 24 | 0 | 2.279908e+04 | 2.707300e+04 | 18.75%| unknown\n", " 29.7s| 1 | 0 | 23243 | - | 676M | 0 | 26k| 20k| 22k|2002 | 3 | 24 | 0 | 2.292005e+04 | 2.707300e+04 | 18.12%| unknown\n", "r29.8s| 1 | 0 | 23243 | - |shifting| 0 | 26k| 20k| 22k|2002 | 3 | 24 | 0 | 2.292005e+04 | 2.515100e+04 | 9.73%| unknown\n", " 30.2s| 1 | 0 | 25780 | - | 680M | 0 | 26k| 20k| 22k|2120 | 4 | 24 | 0 | 2.302305e+04 | 2.515100e+04 | 9.24%| unknown\n", "i32.7s| 1 | 0 | 37528 | - | oneopt| 0 | 26k| 20k| 22k|2120 | 4 | 24 | 0 | 2.302305e+04 | 2.514600e+04 | 9.22%| unknown\n", "(node 1) unresolved numerical troubles in LP 10 -- using pseudo solution instead (loop 1)\n", " 35.8s| 1 | 2 | 49334 | - | 684M | 0 | 26k| 20k| 22k|2120 | 4 | 24 | 0 | 2.302305e+04 | 2.514600e+04 | 9.22%| unknown\n", "d72.5s| 68 | 69 | 98165 | 730.5 |veclendi| 11 | 26k| 20k| 22k| 0 | 1 | 24 |1210 | 2.302805e+04 | 2.511900e+04 | 9.08%| unknown\n", "o82.1s| 92 | 93 |122341 | 803.5 |rootsold| 13 | 26k| 20k| 22k|2233 | 1 | 24 |1568 | 2.302805e+04 | 2.509600e+04 | 8.98%| unknown\n", " 84.6s| 100 | 101 |127795 | 793.7 | 734M | 13 | 26k| 20k| 22k|2251 | 1 | 24 |1659 | 2.302805e+04 | 2.509600e+04 | 8.98%| unknown\n", "d92.0s| 125 | 126 |148749 | 802.7 |guideddi| 15 | 26k| 20k| 22k| 0 | 1 | 24 |1833 | 2.303005e+04 | 2.508900e+04 | 8.94%| unknown\n", "r 114s| 190 | 191 |204186 | 819.9 |ziroundi| 17 | 26k| 20k| 22k|2380 | 1 | 24 |2696 | 2.303205e+04 | 2.508700e+04 | 8.92%| unknown\n", " time | node | left |LP iter|LP it/n|mem/heur|mdpt |vars |cons |rows |cuts |sepa|confs|strbr| dualbound | primalbound | gap | compl. \n", " 118s| 200 | 201 |209863 | 807.3 | 775M | 17 | 26k| 20k| 22k|2406 | 1 | 24 |2831 | 2.303304e+04 | 2.508700e+04 | 8.92%| unknown\n", "r 139s| 262 | 263 |255895 | 791.9 |ziroundi| 20 | 26k| 20k| 22k|2504 | 1 | 28 |3608 | 2.303304e+04 | 2.508500e+04 | 8.91%| unknown\n", "r 139s| 263 | 264 |255962 | 789.1 |ziroundi| 21 | 26k| 20k| 22k|2504 | 1 | 28 |3628 | 2.303304e+04 | 2.508400e+04 | 8.90%| unknown\n", "r 148s| 297 | 298 |269173 | 743.1 |ziroundi| 21 | 26k| 20k| 22k|2538 | 1 | 29 |4061 | 2.303404e+04 | 2.508400e+04 | 8.90%| unknown\n", " 150s| 300 | 301 |274671 | 754.0 | 818M | 21 | 26k| 20k| 22k|2538 | 1 | 30 |4103 | 2.303404e+04 | 2.508400e+04 | 8.90%| unknown\n", "r 154s| 317 | 318 |279506 | 728.8 |ziroundi| 21 | 26k| 20k| 22k|2555 | 1 | 30 |4323 | 2.303404e+04 | 2.508400e+04 | 8.90%| unknown\n", "r 155s| 322 | 323 |279561 | 717.6 |ziroundi| 21 | 26k| 20k| 22k|2555 | 1 | 30 |4365 | 2.303404e+04 | 2.508300e+04 | 8.90%| unknown\n", "r 171s| 390 | 391 |313739 | 680.0 |ziroundi| 21 | 26k| 20k| 22k|2602 | 1 | 35 |5079 | 2.303404e+04 | 2.508300e+04 | 8.90%| unknown\n", " 174s| 400 | 401 |320440 | 679.8 | 881M | 21 | 26k| 20k| 22k|2603 | 1 | 55 |5167 | 2.303404e+04 | 2.508300e+04 | 8.90%| unknown\n", "r 177s| 418 | 419 |326124 | 664.0 |ziroundi| 21 | 26k| 20k| 22k|2603 | 1 | 56 |5295 | 2.303404e+04 | 2.508300e+04 | 8.90%| unknown\n", " 202s| 500 | 501 |405291 | 713.6 | 909M | 21 | 26k| 20k| 22k|2734 | 1 | 65 |5922 | 2.303504e+04 | 2.508300e+04 | 8.89%| unknown\n", " 226s| 600 | 601 |476672 | 713.6 | 922M | 23 | 26k| 20k| 22k|2845 | 2 | 73 |6493 | 2.303504e+04 | 2.508300e+04 | 8.89%| unknown\n", "L 230s| 618 | 619 |481926 | 701.3 | rins| 23 | 26k| 20k| 22k|2853 | 1 | 73 |6520 | 2.303504e+04 | 2.508200e+04 | 8.89%| unknown\n", " 244s| 700 | 701 |519687 | 673.1 | 932M | 23 | 26k| 20k| 22k|2943 | 1 | 77 |6751 | 2.303505e+04 | 2.508200e+04 | 8.89%| unknown\n", " 258s| 800 | 801 |554187 | 632.0 | 936M | 25 | 26k| 20k| 22k|3050 | 2 | 77 |7000 | 2.303602e+04 | 2.508200e+04 | 8.88%| unknown\n", " time | node | left |LP iter|LP it/n|mem/heur|mdpt |vars |cons |rows |cuts |sepa|confs|strbr| dualbound | primalbound | gap | compl. \n", "r 271s| 860 | 861 |603796 | 645.6 |ziroundi| 25 | 26k| 20k| 22k|3093 | 1 | 78 |7085 | 2.303602e+04 | 2.508200e+04*| 8.88%| unknown\n", "r 272s| 874 | 875 |605526 | 637.2 |ziroundi| 25 | 26k| 20k| 22k|3093 | 1 | 78 |7096 | 2.303604e+04 | 2.508000e+04 | 8.87%| unknown\n", "r 272s| 887 | 888 |609332 | 632.2 |ziroundi| 25 | 26k| 20k| 22k|3102 | 1 | 78 |7102 | 2.303604e+04 | 2.508000e+04 | 8.87%| unknown\n", "r 273s| 888 | 889 |611391 | 633.8 |ziroundi| 25 | 26k| 20k| 22k|3102 | 1 | 78 |7102 | 2.303604e+04 | 2.507900e+04 | 8.87%| unknown\n", " 273s| 900 | 901 |613629 | 627.8 | 965M | 25 | 26k| 20k| 22k|3102 | 1 | 78 |7119 | 2.303604e+04 | 2.507900e+04 | 8.87%| unknown\n", " 290s| 1000 | 1001 |659345 | 610.7 | 971M | 25 | 26k| 20k| 22k|3208 | 1 | 81 |7495 | 2.303604e+04 | 2.507900e+04 | 8.87%| unknown\n", "r 298s| 1044 | 1045 |672493 | 597.6 |ziroundi| 27 | 26k| 20k| 22k|3251 | 1 | 81 |7598 | 2.303604e+04 | 2.507800e+04 | 8.86%| unknown\n", "Restart triggered after 50 consecutive estimations that the remaining tree will be large\n", "(run 1, node 1049) performing user restart\n", "\n", "(restart) converted 1743 cuts from the global cut pool into linear constraints\n", "\n", "presolving:\n", "(round 1, exhaustive) 0 del vars, 10 del conss, 0 add conss, 0 chg bounds, 0 chg sides, 0 chg coeffs, 1721 upgd conss, 22621 impls, 1903 clqs\n", "(round 2, medium) 0 del vars, 20 del conss, 10 add conss, 2 chg bounds, 0 chg sides, 2 chg coeffs, 1721 upgd conss, 22626 impls, 1905 clqs\n", "(round 3, exhaustive) 0 del vars, 45 del conss, 10 add conss, 2 chg bounds, 0 chg sides, 2 chg coeffs, 1721 upgd conss, 22626 impls, 1905 clqs\n", "presolving (4 rounds: 4 fast, 4 medium, 3 exhaustive):\n", " 0 deleted vars, 45 deleted constraints, 10 added constraints, 2 tightened bounds, 0 added holes, 0 changed sides, 2 changed coefficients\n", " 22626 implications, 1905 cliques\n", "presolved problem has 26193 variables (17562 bin, 0 int, 0 impl, 8631 cont) and 22285 constraints\n", " 16893 constraints of type \n", " 1599 constraints of type \n", " 3345 constraints of type \n", " 420 constraints of type \n", " 28 constraints of type \n", "transformed objective value is always integral (scale: 1)\n", "Presolving Time: 28.20\n", "transformed 100/100 original solutions to the transformed problem space\n", "\n", "\n", "SCIP Status : solving was interrupted [time limit reached]\n", "Solving Time (sec) : 301.66\n", "Solving Nodes : 0 (total of 1049 nodes in 2 runs)\n", "Primal Bound : +2.50779999999950e+04 (124 solutions)\n", "Dual Bound : +2.30360421900002e+04\n", "Gap : 8.86 %\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "/Users/pablorodriguezmier/miniforge3/envs/corneto/lib/python3.12/site-packages/cvxpy/problems/problem.py:1481: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.\n", " warnings.warn(\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "-------------------------------------------------------------------------------\n", " Summary \n", "-------------------------------------------------------------------------------\n", "(CVXPY) Dec 23 01:26:03 PM: Problem status: optimal_inaccurate\n", "(CVXPY) Dec 23 01:26:03 PM: Optimal value: 2.508e+04\n", "(CVXPY) Dec 23 01:26:03 PM: Compilation took 7.962e+01 seconds\n", "(CVXPY) Dec 23 01:26:03 PM: Solver (including time spent in interface) took 3.072e+02 seconds\n" ] } ], "source": [ "G_dag = G.copy()\n", "new_edges = []\n", "for g in input_nn:\n", " new_edges.append(G_dag.add_edge((), g))\n", "new_edges.append(G_dag.add_edge(\"TCR\", ()))\n", "print(G_dag.shape)\n", "\n", "# Find small DAG. We use Acyclic Flow to find over the space of DAGs\n", "P = cn.opt.AcyclicFlow(G_dag)\n", "# We enforce that the input genes and the output gene are part of the solution\n", "P += P.expr.with_flow[new_edges] == 1\n", "# Minimize the number of active edges\n", "P.add_objectives(sum(P.expr.with_flow), weights=1)\n", "P.solve(solver=\"SCIP\", verbosity=1, max_seconds=300);" ] }, { "cell_type": "code", "execution_count": 37, "metadata": { "editable": true, "slideshow": { "slide_type": "" }, "tags": [] }, "outputs": [ { "data": { "text/plain": [ "((12767, 38388), (12619, 25078))" ] }, "execution_count": 37, "metadata": {}, "output_type": "execute_result" } ], "source": [ "G_subdag = G_dag.edge_subgraph(P.expr.with_flow.value > 0.5)\n", "G_dag.shape, G_subdag.shape" ] }, { "cell_type": "code", "execution_count": 38, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "KPNN edge compression (0-100%): 34.67%\n" ] } ], "source": [ "rel_dag_compression = (1 - (G_subdag.num_edges / G_dag.num_edges)) * 100\n", "print(f\"KPNN edge compression (0-100%): {rel_dag_compression:.2f}%\")" ] }, { "cell_type": "code", "execution_count": 39, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Building DAG NN model with CORNETO using Keras with JAX...\n", " > N. inputs: 12459\n", " > N. outputs: 1\n", " > N. parameters: 12778\n", "Compiling...\n", "Fitting...\n", "Weights saved to /var/folders/b4/gwkwsdb93sv11rtztqbm3l040000gn/T/tmpj5f_9545/pruned_weights_0.keras\n", "\u001b[1m11/11\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m3s\u001b[0m 75ms/step\n", " > Fold 0 validation ROC-AUC=0.992\n", "Building DAG NN model with CORNETO using Keras with JAX...\n", " > N. inputs: 12459\n", " > N. outputs: 1\n", " > N. parameters: 12778\n", "Compiling...\n", "Fitting...\n", "Weights saved to /var/folders/b4/gwkwsdb93sv11rtztqbm3l040000gn/T/tmpj5f_9545/pruned_weights_1.keras\n", "\u001b[1m11/11\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 73ms/step\n", " > Fold 1 validation ROC-AUC=0.983\n", "Building DAG NN model with CORNETO using Keras with JAX...\n", " > N. inputs: 12459\n", " > N. outputs: 1\n", " > N. parameters: 12778\n", "Compiling...\n", "Fitting...\n", "Weights saved to /var/folders/b4/gwkwsdb93sv11rtztqbm3l040000gn/T/tmpj5f_9545/pruned_weights_2.keras\n", "\u001b[1m11/11\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 69ms/step\n", " > Fold 2 validation ROC-AUC=0.987\n", "Building DAG NN model with CORNETO using Keras with JAX...\n", " > N. inputs: 12459\n", " > N. outputs: 1\n", " > N. parameters: 12778\n", "Compiling...\n", "Fitting...\n", "Weights saved to /var/folders/b4/gwkwsdb93sv11rtztqbm3l040000gn/T/tmpj5f_9545/pruned_weights_3.keras\n", "\u001b[1m11/11\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 68ms/step\n", " > Fold 3 validation ROC-AUC=0.994\n", "Building DAG NN model with CORNETO using Keras with JAX...\n", " > N. inputs: 12459\n", " > N. outputs: 1\n", " > N. parameters: 12778\n", "Compiling...\n", "Fitting...\n", "Weights saved to /var/folders/b4/gwkwsdb93sv11rtztqbm3l040000gn/T/tmpj5f_9545/pruned_weights_4.keras\n", "\u001b[1m11/11\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 68ms/step\n", " > Fold 4 validation ROC-AUC=0.994\n", "Validation metrics:\n", " - accuracy: 0.961\n", " - precision: 0.971\n", " - recall: 0.949\n", " - f1: 0.960\n", " - roc_auc: 0.990\n" ] } ], "source": [ "pruned_models, pruned_metrics = stratified_kfold(G_subdag, input_nn, outputs_pkn, file_weights=os.path.join(temp_weights, \"pruned_weights\"))\n", "\n", "print(\"Validation metrics:\")\n", "for k, v in pruned_metrics.items():\n", " print(f\" - {k}: {np.mean(v):.3f}\")" ] }, { "cell_type": "code", "execution_count": 40, "metadata": {}, "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", "
biasabs_biaspow2_bias
foldgene
1SUZ12.EZH2-2.8922452.8922458.365083
4SUZ12.EZH2-2.7365292.7365297.488593
2SUZ12.EZH2-2.5753482.5753486.632418
0SETDB1.NLK.CHD7-2.4760902.4760906.131024
4ZAP70-2.4321732.4321735.915468
3ZAP702.3256422.3256425.408611
0PRKCD-2.3151972.3151975.360137
DUSP3-2.2695322.2695325.150774
4SETDB1.NLK.CHD7-2.2199292.2199294.928085
3NFYA-2.1636132.1636134.681221
\n", "
" ], "text/plain": [ " bias abs_bias pow2_bias\n", "fold gene \n", "1 SUZ12.EZH2 -2.892245 2.892245 8.365083\n", "4 SUZ12.EZH2 -2.736529 2.736529 7.488593\n", "2 SUZ12.EZH2 -2.575348 2.575348 6.632418\n", "0 SETDB1.NLK.CHD7 -2.476090 2.476090 6.131024\n", "4 ZAP70 -2.432173 2.432173 5.915468\n", "3 ZAP70 2.325642 2.325642 5.408611\n", "0 PRKCD -2.315197 2.315197 5.360137\n", " DUSP3 -2.269532 2.269532 5.150774\n", "4 SETDB1.NLK.CHD7 -2.219929 2.219929 4.928085\n", "3 NFYA -2.163613 2.163613 4.681221" ] }, "execution_count": 40, "metadata": {}, "output_type": "execute_result" } ], "source": [ "df_biases_pruned = load_biases(file=os.path.join(temp_weights, \"pruned_weights\"), folds=5)\n", "df_biases_pruned.sort_values(by=\"abs_bias\", ascending=False).head(10)" ] }, { "cell_type": "code", "execution_count": 41, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "gene\n", "SUZ12.EZH2 4.990975\n", "TCR 3.651254\n", "ZAP70 3.080308\n", "SETDB1.NLK.CHD7 2.942616\n", "NFYA 2.748323\n", " ... \n", "PBX1 0.007306\n", "NR2F1 0.005788\n", "ASH2L 0.002699\n", "BCL3 0.002370\n", "GATA3 0.000018\n", "Name: pow2_bias, Length: 160, dtype: float32" ] }, "execution_count": 41, "metadata": {}, "output_type": "execute_result" } ], "source": [ "df_biases_prunedr = df_biases_pruned.copy().reset_index()\n", "gene_biases_score = df_biases_prunedr.groupby(\"gene\")[\"pow2_bias\"].mean().sort_values(ascending=False)\n", "gene_biases_score" ] }, { "cell_type": "code", "execution_count": 42, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "" ] }, "execution_count": 42, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "gene_biases_score.head(30).plot.bar()" ] }, { "cell_type": "code", "execution_count": 43, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Get the top genes sorted by mean bias\n", "top_genes = gene_biases_score.head(30).index\n", "\n", "# Filter the DataFrame to include only rows with these top genes\n", "filtered_df = df_biases_prunedr[df_biases_prunedr['gene'].isin(top_genes)]\n", "\n", "plt.figure(figsize=(10, 6))\n", "sns.violinplot(data=filtered_df, x=\"gene\", y=\"bias\", order=top_genes, dodge=True)\n", "sns.stripplot(data=filtered_df, x=\"gene\", y=\"bias\", order=top_genes, dodge=True)\n", "plt.xticks(rotation=90)\n", "plt.title('Biases of the trained neurons')\n", "plt.xlabel('Gene')\n", "plt.ylabel('Bias')\n", "plt.axhline(0, linestyle=\"--\", color=\"k\")\n", "plt.tight_layout()" ] }, { "cell_type": "code", "execution_count": 44, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Parameter compression: 51.30%\n" ] } ], "source": [ "param_compression = (1-(pruned_models[0].count_params()/models[0].count_params())) * 100\n", "print(f\"Parameter compression: {param_compression:.2f}%\")" ] }, { "cell_type": "code", "execution_count": 45, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Degradation in ROC-AUC after compression (positive = decrease in performance, negative = increase in performance): -0.04%\n" ] } ], "source": [ "perf_degradation = ((np.mean(metrics[\"roc_auc\"]) - np.mean(pruned_metrics[\"roc_auc\"])) / np.mean(metrics[\"roc_auc\"])) * 100\n", "print(f\"Degradation in ROC-AUC after compression (positive = decrease in performance, negative = increase in performance): {perf_degradation:.2f}%\")" ] } ], "metadata": { "colab": { "provenance": [] }, "kernelspec": { "display_name": "Python 3 (ipykernel)", "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": "3.12.7" }, "mystnb": { "execution_mode": "off" } }, "nbformat": 4, "nbformat_minor": 4 }