examples/borzoi_example_sqtl_chr9_135548708_G_C.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "7dbf2734",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-10-06 10:06:41.610527: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:9261] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered\n",
      "2024-10-06 10:06:41.610584: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:607] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered\n",
      "2024-10-06 10:06:41.611618: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1515] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered\n",
      "2024-10-06 10:06:41.618418: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n",
      "To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
      "2024-10-06 10:06:43.358245: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n"
     ]
    }
   ],
   "source": [
    "import json\n",
    "import os\n",
    "import time\n",
    "import warnings\n",
    "\n",
    "import h5py\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.patches as patches\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import pysam\n",
    "import pyfaidx\n",
    "import tensorflow as tf\n",
    "\n",
    "from baskerville import seqnn\n",
    "from baskerville import gene as bgene\n",
    "from baskerville import dna\n",
    "\n",
    "from borzoi_helpers import *\n",
    "\n",
    "tf.compat.v1.logging.set_verbosity(tf.compat.v1.logging.ERROR)\n",
    "#os.environ['CUDA_VISIBLE_DEVICES'] = '-1'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "a4238bad",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "f3c0 model already exists.\n",
      "f3c1 model already exists.\n",
      "f3c2 model already exists.\n",
      "f3c3 model already exists.\n",
      "Gene annotation already exists.\n",
      "Gene annotation (no read-through, protein-coding) already exists.\n",
      "Gene annotation (protein-coding) already exists.\n",
      "TSS annotation already exists.\n",
      "Splice site annotation already exist.\n",
      "Splice site annotation already exist.\n",
      "PolyA site annotation already exist.\n",
      "Human genome FASTA already exists.\n"
     ]
    }
   ],
   "source": [
    "%%bash\n",
    "\n",
    "#Download model weights (data fold 3, 4 replicates)\n",
    "for rep in f3c0,f0 f3c1,f1 f3c2,f2 f3c3,f3; do IFS=\",\"; set -- $rep; \n",
    "  mkdir -p \"saved_models/$1/train\"\n",
    "  local_model=\"saved_models/$1/train/model0_best.h5\"\n",
    "  if [ -f \"$local_model\" ]; then\n",
    "    echo \"$1 model already exists.\"\n",
    "  else\n",
    "    wget --progress=bar:force \"https://storage.googleapis.com/seqnn-share/borzoi/$2/model0_best.h5\" -O \"$local_model\"\n",
    "  fi\n",
    "done\n",
    "\n",
    "#Download and uncompress annotation files\n",
    "mkdir -p hg38/genes/gencode41\n",
    "mkdir -p hg38/genes/polyadb\n",
    "\n",
    "if [ -f hg38/genes/gencode41/gencode41_basic_nort.gtf ]; then\n",
    "  echo \"Gene annotation already exists.\"\n",
    "else\n",
    "  wget -O - https://storage.googleapis.com/seqnn-share/helper/gencode41_basic_nort.gtf.gz | gunzip -c > hg38/genes/gencode41/gencode41_basic_nort.gtf\n",
    "fi\n",
    "\n",
    "if [ -f hg38/genes/gencode41/gencode41_basic_nort_protein.gtf ]; then\n",
    "  echo \"Gene annotation (no read-through, protein-coding) already exists.\"\n",
    "else\n",
    "  wget -O - https://storage.googleapis.com/seqnn-share/helper/gencode41_basic_nort_protein.gtf.gz | gunzip -c > hg38/genes/gencode41/gencode41_basic_nort_protein.gtf\n",
    "fi\n",
    "\n",
    "if [ -f hg38/genes/gencode41/gencode41_basic_protein.gtf ]; then\n",
    "  echo \"Gene annotation (protein-coding) already exists.\"\n",
    "else\n",
    "  wget -O - https://storage.googleapis.com/seqnn-share/helper/gencode41_basic_protein.gtf.gz | gunzip -c > hg38/genes/gencode41/gencode41_basic_protein.gtf\n",
    "fi\n",
    "\n",
    "if [ -f hg38/genes/gencode41/gencode41_basic_tss2.bed ]; then\n",
    "  echo \"TSS annotation already exists.\"\n",
    "else\n",
    "  wget -O - https://storage.googleapis.com/seqnn-share/helper/gencode41_basic_tss2.bed.gz | gunzip -c > hg38/genes/gencode41/gencode41_basic_tss2.bed\n",
    "fi\n",
    "\n",
    "if [ -f hg38/genes/gencode41/gencode41_basic_protein_splice.csv.gz ]; then\n",
    "  echo \"Splice site annotation already exist.\"\n",
    "else\n",
    "  wget https://storage.googleapis.com/seqnn-share/helper/gencode41_basic_protein_splice.csv.gz -O hg38/genes/gencode41/gencode41_basic_protein_splice.csv.gz\n",
    "fi\n",
    "\n",
    "if [ -f hg38/genes/gencode41/gencode41_basic_protein_splice.gff ]; then\n",
    "  echo \"Splice site annotation already exist.\"\n",
    "else\n",
    "  wget -O - https://storage.googleapis.com/seqnn-share/helper/gencode41_basic_protein_splice.gff.gz | gunzip -c > hg38/genes/gencode41/gencode41_basic_protein_splice.gff\n",
    "fi\n",
    "\n",
    "if [ -f hg38/genes/polyadb/polyadb_human_v3.csv.gz ]; then\n",
    "  echo \"PolyA site annotation already exist.\"\n",
    "else\n",
    "  wget https://storage.googleapis.com/seqnn-share/helper/polyadb_human_v3.csv.gz -O hg38/genes/polyadb/polyadb_human_v3.csv.gz\n",
    "fi\n",
    "\n",
    "#Download and index hg38 genome\n",
    "mkdir -p hg38/assembly/ucsc\n",
    "\n",
    "if [ -f hg38/assembly/ucsc/hg38.fa ]; then\n",
    "  echo \"Human genome FASTA already exists.\"\n",
    "else\n",
    "  wget -O - http://hgdownload.cse.ucsc.edu/goldenPath/hg38/bigZips/hg38.fa.gz | gunzip -c > hg38/assembly/ucsc/hg38.fa\n",
    "fi\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "c4d7b5af",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Faidx(\"hg38/assembly/ucsc/hg38.fa\")"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pyfaidx.Faidx('hg38/assembly/ucsc/hg38.fa')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "afa277c3",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-10-06 10:07:00.831023: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1929] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 10520 MB memory:  -> device: 0, name: NVIDIA GeForce GTX 1080 Ti, pci bus id: 0000:02:00.0, compute capability: 6.1\n"
     ]
    }
   ],
   "source": [
    "#Model configuration\n",
    "\n",
    "params_file = 'params_pred.json'\n",
    "targets_file = 'targets_gtex.txt' #Subset of targets_human.txt\n",
    "\n",
    "seq_len = 524288\n",
    "n_reps = 1       #To use only one model replicate, set to 'n_reps = 1'. To use all four replicates, set 'n_reps = 4'.\n",
    "rc = True         #Average across reverse-complement prediction\n",
    "\n",
    "#Read model parameters\n",
    "\n",
    "with open(params_file) as params_open :\n",
    "    \n",
    "    params = json.load(params_open)\n",
    "    \n",
    "    params_model = params['model']\n",
    "    params_train = params['train']\n",
    "\n",
    "#Read targets\n",
    "\n",
    "targets_df = pd.read_csv(targets_file, index_col=0, sep='\\t')\n",
    "target_index = targets_df.index\n",
    "\n",
    "#Create local index of strand_pair (relative to sliced targets)\n",
    "if rc :\n",
    "    strand_pair = targets_df.strand_pair\n",
    "    \n",
    "    target_slice_dict = {ix : i for i, ix in enumerate(target_index.values.tolist())}\n",
    "    slice_pair = np.array([\n",
    "        target_slice_dict[ix] if ix in target_slice_dict else ix for ix in strand_pair.values.tolist()\n",
    "    ], dtype='int32')\n",
    "\n",
    "#Initialize model ensemble\n",
    "\n",
    "models = []\n",
    "for rep_ix in range(n_reps) :\n",
    "    \n",
    "    model_file = \"saved_models/f3c\" + str(rep_ix) + \"/train/model0_best.h5\"\n",
    "\n",
    "    seqnn_model = seqnn.SeqNN(params_model)\n",
    "    seqnn_model.restore(model_file, 0)\n",
    "    seqnn_model.build_slice(target_index)\n",
    "    if rc :\n",
    "        seqnn_model.strand_pair.append(slice_pair)\n",
    "    seqnn_model.build_ensemble(rc, [0])\n",
    "    \n",
    "    models.append(seqnn_model)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "505c8f56",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "len(splice_df) = 404837\n"
     ]
    }
   ],
   "source": [
    "#Initialize fasta sequence extractor\n",
    "\n",
    "fasta_open = pysam.Fastafile('hg38/assembly/ucsc/hg38.fa')\n",
    "\n",
    "#Load splice site annotation\n",
    "\n",
    "splice_df = pd.read_csv('hg38/genes/gencode41/gencode41_basic_protein_splice.csv.gz', sep='\\t', compression='gzip')\n",
    "\n",
    "print(\"len(splice_df) = \" + str(len(splice_df)))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "2c3584e4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "testis tracks = [79, 80]\n"
     ]
    }
   ],
   "source": [
    "#Print index of GTEx testis tracks in targets file\n",
    "\n",
    "targets_df['local_index'] = np.arange(len(targets_df))\n",
    "\n",
    "print(\"testis tracks = \" + str(targets_df.loc[targets_df['description'] == 'RNA:testis']['local_index'].tolist()))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "42154bfe",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-10-06 10:08:03.084851: I external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:454] Loaded cuDNN version 8907\n",
      "2024-10-06 10:08:03.154520: I external/local_tsl/tsl/platform/default/subprocess.cc:304] Start cannot spawn child process: No such file or directory\n",
      "2024-10-06 10:08:03.472584: I external/local_tsl/tsl/platform/default/subprocess.cc:304] Start cannot spawn child process: No such file or directory\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-- Counts --\n",
      " - max_y_wt = 60.267\n",
      " - max_y_mut = 60.4653\n",
      " -- (max_y = 60.4653)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " - max_y_wt = 361.0789\n",
      " - max_y_mut = 360.6328\n",
      " -- (max_y = 361.0789)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 17.6 s, sys: 593 ms, total: 18.2 s\n",
      "Wall time: 21.1 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "#Predict for chr9_135548708_G_C\n",
    "# (~6 minutes on CPU w 1 replicate; ~15 seconds on GPU)\n",
    "\n",
    "save_figs = False\n",
    "save_suffix = '_chr9_135548708_G_C'\n",
    "\n",
    "center_pos = 135548708\n",
    "\n",
    "chrom = 'chr9'\n",
    "poses = [135548708]\n",
    "alts = ['C']\n",
    "\n",
    "start = center_pos - seq_len // 2\n",
    "end = center_pos + seq_len // 2\n",
    "\n",
    "sequence_one_hot_wt = process_sequence(fasta_open, chrom, start, end)\n",
    "\n",
    "#Induce mutation(s)\n",
    "sequence_one_hot_mut = np.copy(sequence_one_hot_wt)\n",
    "\n",
    "for pos, alt in zip(poses, alts) :\n",
    "    alt_ix = -1\n",
    "    if alt == 'A' :\n",
    "        alt_ix = 0\n",
    "    elif alt == 'C' :\n",
    "        alt_ix = 1\n",
    "    elif alt == 'G' :\n",
    "        alt_ix = 2\n",
    "    elif alt == 'T' :\n",
    "        alt_ix = 3\n",
    "\n",
    "    sequence_one_hot_mut[pos-start-1] = 0.\n",
    "    sequence_one_hot_mut[pos-start-1, alt_ix] = 1.\n",
    "\n",
    "#Make predictions\n",
    "y_wt = predict_tracks(models, sequence_one_hot_wt)\n",
    "y_mut = predict_tracks(models, sequence_one_hot_mut)\n",
    "\n",
    "#Visualize quantized tracks over SNP\n",
    "\n",
    "plot_window = 4096\n",
    "bin_size = 32\n",
    "pad = 16\n",
    "\n",
    "untransform_old = True\n",
    "normalize_counts = False\n",
    "\n",
    "anno_df = splice_df\n",
    "\n",
    "#Tracks\n",
    "track_indices = [\n",
    "    np.arange(0, 89).tolist(),\n",
    "    [79, 80],\n",
    "]\n",
    "\n",
    "track_names = [\n",
    "    'GTEx Coverage (All tissues)',\n",
    "    'GTEx Coverage (Testis)',\n",
    "]\n",
    "\n",
    "track_scales = [0.01]*2\n",
    "track_transforms = [3./4.]*2\n",
    "soft_clips = [384.]*2\n",
    "\n",
    "print(\"-- Counts --\")\n",
    "plot_coverage_track_pair_bins(\n",
    "    y_wt,\n",
    "    y_mut,\n",
    "    chrom,\n",
    "    start,\n",
    "    center_pos,\n",
    "    poses,\n",
    "    track_indices,\n",
    "    track_names,\n",
    "    track_scales,\n",
    "    track_transforms,\n",
    "    soft_clips,\n",
    "    plot_window=plot_window,\n",
    "    normalize_window=8*plot_window,\n",
    "    bin_size=bin_size,\n",
    "    pad=pad,\n",
    "    normalize_counts=normalize_counts,\n",
    "    save_figs=save_figs,\n",
    "    save_suffix=save_suffix,\n",
    "    anno_df=anno_df,\n",
    "    untransform_old=untransform_old,\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "f3bb0e25",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mut_bin = 8176\n",
      "\n",
      "splice_ix = 0: bin = 8115 (8113 - 8118)\n",
      "splice_ix = 1: bin = 8119 (8117 - 8122)\n",
      "splice_ix = 2: bin = 8128 (8126 - 8131)\n",
      "splice_ix = 3: bin = 8130 (8128 - 8133)\n",
      "splice_ix = 4: bin = 8149 (8147 - 8152)\n",
      "splice_ix = 5: bin = 8153 (8151 - 8156)\n",
      "splice_ix = 6: bin = 8173 (8171 - 8176)\n",
      "splice_ix = 7: bin = 8175 (8173 - 8178)\n",
      "splice_ix = 8: bin = 8179 (8177 - 8182)\n",
      "splice_ix = 9: bin = 8194 (8192 - 8197)\n",
      "splice_ix = 10: bin = 8195 (8193 - 8198)\n",
      "splice_ix = 11: bin = 8195 (8193 - 8198)\n",
      "splice_ix = 12: bin = 8202 (8200 - 8205)\n",
      "splice_ix = 13: bin = 8211 (8209 - 8214)\n",
      "splice_ix = 14: bin = 8211 (8209 - 8214)\n"
     ]
    }
   ],
   "source": [
    "#Print splice site bin positions and other info\n",
    "\n",
    "plot_start = center_pos - plot_window // 2\n",
    "plot_end = center_pos + plot_window // 2\n",
    "\n",
    "plot_start_bin = (plot_start - start) // bin_size - pad\n",
    "plot_end_bin = (plot_end - start) // bin_size - pad\n",
    "\n",
    "mut_bin = (poses[0] - start) // bin_size - pad\n",
    "\n",
    "#Get splice site positions\n",
    "splice_poses = splice_df.query(\"chrom == '\" + chrom + \"' and position_hg38 >= \" + str(plot_start) + \" and position_hg38 < \" + str(plot_end))['position_hg38'].unique().tolist()\n",
    "\n",
    "print(\"mut_bin = \" + str(mut_bin))\n",
    "print(\"\")\n",
    "\n",
    "#Print splice site positions\n",
    "for splice_ix, splice_pos in enumerate(splice_poses) :\n",
    "    \n",
    "    splice_bin = int((splice_pos - start) // 32) - 16\n",
    "    \n",
    "    bin_end = splice_bin + 3\n",
    "    bin_start = bin_end - 5\n",
    "    \n",
    "    print(\"splice_ix = \" + str(splice_ix) + \": bin = \" + str(splice_bin) + \" (\" + str(bin_start) + \" - \" + str(bin_end) + \")\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "b9b9f717",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Choose splice sites to use for attributions\n",
    "\n",
    "exon_1_start_bin = 8173\n",
    "exon_1_end_bin = 8176\n",
    "\n",
    "exon_2_start_bin = 8180\n",
    "exon_2_end_bin = 8183\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "41016527",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 414s 414s/step\n",
      "1/1 [==============================] - 220s 220s/step\n",
      "1/1 [==============================] - 409s 409s/step\n",
      "1/1 [==============================] - 220s 220s/step\n",
      "CPU times: user 18min 44s, sys: 2min 45s, total: 21min 30s\n",
      "Wall time: 22min 55s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "#Get contribution scores (gradient) for testis GTEX tracks\n",
    "# (~20 minutes on CPU; otherwise runnable only on 40GB GPU cards, e.g. A100)\n",
    "\n",
    "_, _, [pred_grad_wt, pred_grad_mut] = get_prediction_gradient_w_rc(\n",
    "    models,\n",
    "    [sequence_one_hot_wt, sequence_one_hot_mut],\n",
    "    exon_2_start_bin,\n",
    "    exon_2_end_bin,\n",
    "    exon_1_start_bin,\n",
    "    exon_1_end_bin,\n",
    "    target_index[[79, 80]].tolist(),\n",
    "    0.01,\n",
    "    3./4.,\n",
    "    clip_soft=384.,\n",
    "    use_mean=False,\n",
    "    use_ratio=True,\n",
    "    use_logodds=False,\n",
    "    subtract_avg=True,\n",
    "    fold_index=np.arange(n_reps).tolist(),\n",
    "    untransform_old=True,\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "414eb8be",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--- WT ---\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--- Mut ---\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Visualize gradient contribution scores (128 bp window centered on variant)\n",
    "\n",
    "visualize_input_gradient_pair(\n",
    "    pred_grad_wt,\n",
    "    pred_grad_mut,\n",
    "    plot_start=(poses[0] - start) - 64,\n",
    "    plot_end=(poses[0] - start) + 64,\n",
    "    save_figs=False,\n",
    "    fig_name=chrom + '_' + str(poses[0]) + '_prediction_grad_gtex_testis'\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "7f2bd42d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "example_ix = 0\n",
      "example_ix = 1\n",
      "CPU times: user 15min 37s, sys: 1min 15s, total: 16min 53s\n",
      "Wall time: 33min 18s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "#Get contribution scores (ISM) for testis GTEX tracks\n",
    "# (not feasible to run on CPU; ~30 minutes on GPU)\n",
    "\n",
    "[pred_ism_wt, pred_ism_mut] = get_ism(\n",
    "    models,\n",
    "    [sequence_one_hot_wt, sequence_one_hot_mut],\n",
    "    (poses[0] - start) - 64,\n",
    "    (poses[0] - start) + 64,\n",
    "    exon_2_start_bin,\n",
    "    exon_2_end_bin,\n",
    "    exon_1_start_bin,\n",
    "    exon_1_end_bin,\n",
    "    [79, 80],\n",
    "    0.01,\n",
    "    3./4.,\n",
    "    384.,\n",
    "    use_mean=False,\n",
    "    use_ratio=True,\n",
    "    use_logodds=False,\n",
    "    untransform_old=True,\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "3bd730e9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--- WT ---\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--- Mut ---\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Visualize ISM contribution scores (128 bp window centered on variant)\n",
    "\n",
    "visualize_input_gradient_pair(\n",
    "    pred_ism_wt,\n",
    "    pred_ism_mut,\n",
    "    plot_start=(poses[0] - start) - 64,\n",
    "    plot_end=(poses[0] - start) + 64,\n",
    "    save_figs=False,\n",
    "    fig_name=chrom + '_' + str(poses[0]) + '_prediction_ism_gtex_testis'\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "0132c170",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "example_ix = 0\n",
      "example_ix = 1\n",
      "CPU times: user 1h 2min, sys: 5min 17s, total: 1h 7min 17s\n",
      "Wall time: 2h 13min 1s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "#Get contribution scores (ISM shuffle) for testis GTEX tracks\n",
    "# (not feasible to run on CPU; ~135 minutes on GPU)\n",
    "\n",
    "[pred_ism_wt, pred_ism_mut] = get_ism_shuffle(\n",
    "    models,\n",
    "    [sequence_one_hot_wt, sequence_one_hot_mut],\n",
    "    (poses[0] - start) - 64,\n",
    "    (poses[0] - start) + 64,\n",
    "    exon_2_start_bin,\n",
    "    exon_2_end_bin,\n",
    "    exon_1_start_bin,\n",
    "    exon_1_end_bin,\n",
    "    [79, 80],\n",
    "    0.01,\n",
    "    3./4.,\n",
    "    384.,\n",
    "    window_size=5,\n",
    "    n_samples=12,\n",
    "    use_mean=False,\n",
    "    use_ratio=True,\n",
    "    use_logodds=False,\n",
    "    untransform_old=True,\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "34d5e490",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--- WT ---\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--- Mut ---\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Visualize ISM Shuffle contribution scores (128 bp window centered on variant)\n",
    "\n",
    "visualize_input_gradient_pair(\n",
    "    pred_ism_wt,\n",
    "    pred_ism_mut,\n",
    "    plot_start=(poses[0] - start) - 64,\n",
    "    plot_end=(poses[0] - start) + 64,\n",
    "    save_figs=False,\n",
    "    fig_name=chrom + '_' + str(poses[0]) + '_prediction_ism_shuffle_gtex_testis'\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "5a067915",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Load samples with reference- and alternate alleles respectively\n",
    "\n",
    "#These files are protected by dbGaP - email to request access\n",
    "\n",
    "cov_files_wt = pd.read_csv(\"gtex_ref_chr9_135548708_G_C.txt\", sep='\\t', names=['file'])['file'].values.tolist()\n",
    "cov_files_mut = pd.read_csv(\"gtex_alt_chr9_135548708_G_C.txt\", sep='\\t', names=['file'])['file'].values.tolist()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "6d9948c8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "gtex_targets_wt.shape = (16352, 5)\n",
      "gtex_targets_mut.shape = (16352, 5)\n",
      "-- Counts --\n",
      " - max_y_wt = 986.4816\n",
      " - max_y_mut = 1083.784\n",
      " -- (max_y = 1083.784)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Read measured coverage tracks for chr9_135548708_G_C\n",
    "\n",
    "save_figs = False\n",
    "save_suffix = '_chr9_135548708_G_C_meas_5_subjects'\n",
    "\n",
    "center_pos = 135548708\n",
    "\n",
    "chrom = 'chr9'\n",
    "poses = [135548708]\n",
    "alts = ['C']\n",
    "\n",
    "start = center_pos - seq_len // 2\n",
    "end = center_pos + seq_len // 2\n",
    "\n",
    "blacklist_bed = \"hg38/blacklist/blacklist_hg38_all.bed\"\n",
    "\n",
    "read_coverage_func_wt, close_coverage_func_wt = get_coverage_reader(cov_files_wt, 16384, 16, blacklist_bed)\n",
    "read_coverage_func_mut, close_coverage_func_mut = get_coverage_reader(cov_files_mut, 16384, 16, blacklist_bed)\n",
    "\n",
    "gtex_targets_wt = read_coverage_func_wt(chrom, start, end, clip_soft=384., clip=768., scale=0.01, transform_old=True)\n",
    "gtex_targets_mut = read_coverage_func_mut(chrom, start, end, clip_soft=384., clip=768., scale=0.01, transform_old=True)\n",
    "\n",
    "close_coverage_func_wt()\n",
    "close_coverage_func_mut()\n",
    "\n",
    "print(\"gtex_targets_wt.shape = \" + str(gtex_targets_wt.shape))\n",
    "print(\"gtex_targets_mut.shape = \" + str(gtex_targets_mut.shape))\n",
    "\n",
    "plot_window = 4096\n",
    "bin_size = 32\n",
    "pad = 16\n",
    "\n",
    "anno_df = splice_df\n",
    "\n",
    "untransform_old = True\n",
    "normalize_counts = True\n",
    "\n",
    "#Tracks\n",
    "track_indices = [\n",
    "    [0, 1, 2, 3, 4],\n",
    "]\n",
    "\n",
    "track_names = [\n",
    "    'GTEx Testis measured (5 subjects)',\n",
    "]\n",
    "\n",
    "track_scales = [\n",
    "    0.01,\n",
    "]\n",
    "\n",
    "track_transforms = [\n",
    "    3./4.,\n",
    "]\n",
    "\n",
    "soft_clips = [\n",
    "    384.,\n",
    "]\n",
    "\n",
    "print(\"-- Counts --\")\n",
    "plot_coverage_track_pair_bins(\n",
    "    gtex_targets_wt[None, None, ...],\n",
    "    gtex_targets_mut[None, None, ...],\n",
    "    chrom,\n",
    "    start,\n",
    "    center_pos,\n",
    "    poses,\n",
    "    track_indices,\n",
    "    track_names,\n",
    "    track_scales,\n",
    "    track_transforms,\n",
    "    soft_clips,\n",
    "    plot_window=plot_window,\n",
    "    normalize_window=8*plot_window,\n",
    "    bin_size=bin_size,\n",
    "    pad=pad,\n",
    "    normalize_counts=normalize_counts,\n",
    "    save_figs=save_figs,\n",
    "    save_suffix=save_suffix,\n",
    "    anno_df=anno_df,\n",
    "    untransform_old=untransform_old,\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7efea964",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "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": "3.8.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}