{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "265129ef-a7d8-44b2-a466-e93925f11860",
   "metadata": {},
   "source": [
    "# Higher-order interaction test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "0e520279-ba5f-4872-a38d-27487ed26c56",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-06-10 15:32:45.740168: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  SSE4.1 SSE4.2 AVX AVX2 FMA\n",
      "To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\n"
     ]
    }
   ],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "from creme import creme\n",
    "from creme import utils\n",
    "from creme import custom_model\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import seaborn as sns\n",
    "import shuffle"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f3083d60-240d-4c05-afc2-02bded899c81",
   "metadata": {},
   "source": [
    "Load Enformer and example sequences"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "5e3979fd-a872-41fa-afc4-109ee830b141",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-06-10 15:32:47.349063: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  SSE4.1 SSE4.2 AVX AVX2 FMA\n",
      "To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
      "2024-06-10 15:32:48.819795: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1613] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 67919 MB memory:  -> device: 0, name: NVIDIA A100 80GB PCIe, pci bus id: 0000:85:00.0, compute capability: 8.0\n"
     ]
    }
   ],
   "source": [
    "data_dir = '../../../data'\n",
    "track_index = [5111]\n",
    "# TSS bin indeces\n",
    "bins = [447, 448]\n",
    "model = custom_model.Enformer(track_index=track_index, bin_index=bins)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "5b84bb66-1c29-448f-b78f-3ec690e39d22",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-06-10 15:32:53.943122: I tensorflow/compiler/xla/stream_executor/cuda/cuda_blas.cc:630] TensorFloat-32 will be used for the matrix multiplication. This will only be logged once.\n",
      "2024-06-10 15:32:54.196676: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:428] Loaded cuDNN version 8401\n"
     ]
    }
   ],
   "source": [
    "fasta_path = f'{data_dir}/GRCh38.primary_assembly.genome.fa'\n",
    "seq_parser = utils.SequenceParser(fasta_path)\n",
    "\n",
    "gene = 'CLUAP1_chr16_3501011_+'\n",
    "\n",
    "gene_name, chrom, start, strand = gene.split('_')\n",
    "wt_seq = seq_parser.extract_seq_centered(chrom, int(start), strand, model.seq_length)\n",
    "wt = model.predict(wt_seq)[0,:,0].mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "727193db-efab-48b9-9fe8-7fd4ed36b89a",
   "metadata": {},
   "source": [
    "**Higher-order interaction test**\n",
    "\n",
    "\n",
    "In this example we will comprehensively analyze the context of the GATA2 gene using a greedy search for necessary tiles.\n",
    "\n",
    "To run the test we need:\n",
    "- a loaded model\n",
    "- a list of tile coordinates to test\n",
    "- optimization - choice between [np.argmin, np.argmax] which determines if we will search for enhancers or silencers\n",
    "- num_shuffle - number of shuffled\n",
    "- number of iterations or tiles to shuffle\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "fa851a1f-f266-4142-821b-333d08aa684b",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "perturb_window = 5000\n",
    "N_shuffles = 10\n",
    "tss_tile, cre_tiles = utils.set_tile_range(model.seq_length, perturb_window)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "a2913af0-705d-46c3-8b75-6389b2344cdf",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "38"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(cre_tiles)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b0193ee8-b6d3-4e89-ad06-7448ae91ef62",
   "metadata": {},
   "source": [
    "CLUAP1 gene context"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "52086ff5-48a3-45b2-99ad-818cd054e529",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|                                                                                     | 0/3 [00:00<?, ?it/s]\n",
      "  0%|                                                                                    | 0/38 [00:00<?, ?it/s]\u001b[A\n",
      "  3%|██                                                                          | 1/38 [00:03<01:54,  3.10s/it]\u001b[A\n",
      "  5%|████                                                                        | 2/38 [00:06<01:51,  3.10s/it]\u001b[A\n",
      "  8%|██████                                                                      | 3/38 [00:09<01:48,  3.10s/it]\u001b[A\n",
      " 11%|████████                                                                    | 4/38 [00:12<01:45,  3.10s/it]\u001b[A\n",
      " 13%|██████████                                                                  | 5/38 [00:15<01:42,  3.10s/it]\u001b[A\n",
      " 16%|████████████                                                                | 6/38 [00:18<01:39,  3.10s/it]\u001b[A\n",
      " 18%|██████████████                                                              | 7/38 [00:21<01:36,  3.11s/it]\u001b[A\n",
      " 21%|████████████████                                                            | 8/38 [00:24<01:33,  3.10s/it]\u001b[A\n",
      " 24%|██████████████████                                                          | 9/38 [00:27<01:29,  3.10s/it]\u001b[A\n",
      " 26%|███████████████████▋                                                       | 10/38 [00:30<01:26,  3.10s/it]\u001b[A\n",
      " 29%|█████████████████████▋                                                     | 11/38 [00:34<01:23,  3.10s/it]\u001b[A\n",
      " 32%|███████████████████████▋                                                   | 12/38 [00:37<01:20,  3.11s/it]\u001b[A\n",
      " 34%|█████████████████████████▋                                                 | 13/38 [00:40<01:17,  3.12s/it]\u001b[A\n",
      " 37%|███████████████████████████▋                                               | 14/38 [00:43<01:14,  3.12s/it]\u001b[A\n",
      " 39%|█████████████████████████████▌                                             | 15/38 [00:46<01:11,  3.12s/it]\u001b[A\n",
      " 42%|███████████████████████████████▌                                           | 16/38 [00:49<01:08,  3.13s/it]\u001b[A\n",
      " 45%|█████████████████████████████████▌                                         | 17/38 [00:52<01:05,  3.13s/it]\u001b[A\n",
      " 47%|███████████████████████████████████▌                                       | 18/38 [00:56<01:02,  3.13s/it]\u001b[A\n",
      " 50%|█████████████████████████████████████▌                                     | 19/38 [00:59<00:59,  3.13s/it]\u001b[A\n",
      " 53%|███████████████████████████████████████▍                                   | 20/38 [01:02<00:56,  3.13s/it]\u001b[A\n",
      " 55%|█████████████████████████████████████████▍                                 | 21/38 [01:05<00:53,  3.13s/it]\u001b[A\n",
      " 58%|███████████████████████████████████████████▍                               | 22/38 [01:08<00:50,  3.13s/it]\u001b[A\n",
      " 61%|█████████████████████████████████████████████▍                             | 23/38 [01:11<00:46,  3.13s/it]\u001b[A\n",
      " 63%|███████████████████████████████████████████████▎                           | 24/38 [01:14<00:43,  3.13s/it]\u001b[A\n",
      " 66%|█████████████████████████████████████████████████▎                         | 25/38 [01:17<00:40,  3.13s/it]\u001b[A\n",
      " 68%|███████████████████████████████████████████████████▎                       | 26/38 [01:21<00:37,  3.13s/it]\u001b[A\n",
      " 71%|█████████████████████████████████████████████████████▎                     | 27/38 [01:24<00:34,  3.13s/it]\u001b[A\n",
      " 74%|███████████████████████████████████████████████████████▎                   | 28/38 [01:27<00:31,  3.13s/it]\u001b[A\n",
      " 76%|█████████████████████████████████████████████████████████▏                 | 29/38 [01:30<00:28,  3.14s/it]\u001b[A\n",
      " 79%|███████████████████████████████████████████████████████████▏               | 30/38 [01:33<00:25,  3.14s/it]\u001b[A\n",
      " 82%|█████████████████████████████████████████████████████████████▏             | 31/38 [01:36<00:21,  3.14s/it]\u001b[A\n",
      " 84%|███████████████████████████████████████████████████████████████▏           | 32/38 [01:39<00:18,  3.14s/it]\u001b[A\n",
      " 87%|█████████████████████████████████████████████████████████████████▏         | 33/38 [01:43<00:15,  3.14s/it]\u001b[A\n",
      " 89%|███████████████████████████████████████████████████████████████████        | 34/38 [01:46<00:12,  3.14s/it]\u001b[A\n",
      " 92%|█████████████████████████████████████████████████████████████████████      | 35/38 [01:49<00:09,  3.14s/it]\u001b[A\n",
      " 95%|███████████████████████████████████████████████████████████████████████    | 36/38 [01:52<00:06,  3.14s/it]\u001b[A\n",
      " 97%|█████████████████████████████████████████████████████████████████████████  | 37/38 [01:55<00:03,  3.14s/it]\u001b[A\n",
      "100%|███████████████████████████████████████████████████████████████████████████| 38/38 [01:58<00:00,  3.13s/it]\u001b[A\n",
      " 33%|█████████████████████████▎                                                  | 1/3 [01:59<03:58, 119.38s/it]\n",
      "  0%|                                                                                    | 0/37 [00:00<?, ?it/s]\u001b[A\n",
      "  3%|██                                                                          | 1/37 [00:03<01:52,  3.13s/it]\u001b[A\n",
      "  5%|████                                                                        | 2/37 [00:06<01:49,  3.13s/it]\u001b[A\n",
      "  8%|██████▏                                                                     | 3/37 [00:09<01:46,  3.13s/it]\u001b[A\n",
      " 11%|████████▏                                                                   | 4/37 [00:12<01:43,  3.13s/it]\u001b[A\n",
      " 14%|██████████▎                                                                 | 5/37 [00:15<01:40,  3.13s/it]\u001b[A\n",
      " 16%|████████████▎                                                               | 6/37 [00:18<01:36,  3.13s/it]\u001b[A\n",
      " 19%|██████████████▍                                                             | 7/37 [00:21<01:33,  3.13s/it]\u001b[A\n",
      " 22%|████████████████▍                                                           | 8/37 [00:25<01:30,  3.13s/it]\u001b[A\n",
      " 24%|██████████████████▍                                                         | 9/37 [00:28<01:27,  3.13s/it]\u001b[A\n",
      " 27%|████████████████████▎                                                      | 10/37 [00:31<01:24,  3.13s/it]\u001b[A\n",
      " 30%|██████████████████████▎                                                    | 11/37 [00:34<01:21,  3.13s/it]\u001b[A\n",
      " 32%|████████████████████████▎                                                  | 12/37 [00:37<01:18,  3.13s/it]\u001b[A\n",
      " 35%|██████████████████████████▎                                                | 13/37 [00:40<01:15,  3.13s/it]\u001b[A\n",
      " 38%|████████████████████████████▍                                              | 14/37 [00:43<01:11,  3.13s/it]\u001b[A\n",
      " 41%|██████████████████████████████▍                                            | 15/37 [00:46<01:08,  3.13s/it]\u001b[A\n",
      " 43%|████████████████████████████████▍                                          | 16/37 [00:50<01:05,  3.13s/it]\u001b[A\n",
      " 46%|██████████████████████████████████▍                                        | 17/37 [00:53<01:02,  3.13s/it]\u001b[A\n",
      " 49%|████████████████████████████████████▍                                      | 18/37 [00:56<00:59,  3.13s/it]\u001b[A\n",
      " 51%|██████████████████████████████████████▌                                    | 19/37 [00:59<00:56,  3.13s/it]\u001b[A\n",
      " 54%|████████████████████████████████████████▌                                  | 20/37 [01:02<00:53,  3.13s/it]\u001b[A\n",
      " 57%|██████████████████████████████████████████▌                                | 21/37 [01:05<00:50,  3.13s/it]\u001b[A\n",
      " 59%|████████████████████████████████████████████▌                              | 22/37 [01:08<00:46,  3.13s/it]\u001b[A\n",
      " 62%|██████████████████████████████████████████████▌                            | 23/37 [01:12<00:43,  3.14s/it]\u001b[A\n",
      " 65%|████████████████████████████████████████████████▋                          | 24/37 [01:15<00:40,  3.14s/it]\u001b[A\n",
      " 68%|██████████████████████████████████████████████████▋                        | 25/37 [01:18<00:37,  3.14s/it]\u001b[A\n",
      " 70%|████████████████████████████████████████████████████▋                      | 26/37 [01:21<00:34,  3.14s/it]\u001b[A\n",
      " 73%|██████████████████████████████████████████████████████▋                    | 27/37 [01:24<00:31,  3.14s/it]\u001b[A\n",
      " 76%|████████████████████████████████████████████████████████▊                  | 28/37 [01:27<00:28,  3.14s/it]\u001b[A\n",
      " 78%|██████████████████████████████████████████████████████████▊                | 29/37 [01:30<00:25,  3.14s/it]\u001b[A\n",
      " 81%|████████████████████████████████████████████████████████████▊              | 30/37 [01:34<00:22,  3.14s/it]\u001b[A\n",
      " 84%|██████████████████████████████████████████████████████████████▊            | 31/37 [01:37<00:18,  3.14s/it]\u001b[A\n",
      " 86%|████████████████████████████████████████████████████████████████▊          | 32/37 [01:40<00:15,  3.14s/it]\u001b[A\n",
      " 89%|██████████████████████████████████████████████████████████████████▉        | 33/37 [01:43<00:12,  3.15s/it]\u001b[A\n",
      " 92%|████████████████████████████████████████████████████████████████████▉      | 34/37 [01:46<00:09,  3.15s/it]\u001b[A\n",
      " 95%|██████████████████████████████████████████████████████████████████████▉    | 35/37 [01:49<00:06,  3.15s/it]\u001b[A\n",
      " 97%|████████████████████████████████████████████████████████████████████████▉  | 36/37 [01:52<00:03,  3.15s/it]\u001b[A\n",
      "100%|███████████████████████████████████████████████████████████████████████████| 37/37 [01:56<00:00,  3.14s/it]\u001b[A\n",
      " 67%|██████████████████████████████████████████████████▋                         | 2/3 [03:56<01:57, 117.78s/it]\n",
      "  0%|                                                                                    | 0/36 [00:00<?, ?it/s]\u001b[A\n",
      "  3%|██                                                                          | 1/36 [00:03<01:49,  3.14s/it]\u001b[A\n",
      "  6%|████▏                                                                       | 2/36 [00:06<01:47,  3.15s/it]\u001b[A\n",
      "  8%|██████▎                                                                     | 3/36 [00:09<01:43,  3.15s/it]\u001b[A\n",
      " 11%|████████▍                                                                   | 4/36 [00:12<01:40,  3.15s/it]\u001b[A\n",
      " 14%|██████████▌                                                                 | 5/36 [00:15<01:37,  3.15s/it]\u001b[A\n",
      " 17%|████████████▋                                                               | 6/36 [00:18<01:34,  3.15s/it]\u001b[A\n",
      " 19%|██████████████▊                                                             | 7/36 [00:22<01:31,  3.15s/it]\u001b[A\n",
      " 22%|████████████████▉                                                           | 8/36 [00:25<01:28,  3.15s/it]\u001b[A\n",
      " 25%|███████████████████                                                         | 9/36 [00:28<01:24,  3.14s/it]\u001b[A\n",
      " 28%|████████████████████▊                                                      | 10/36 [00:31<01:22,  3.15s/it]\u001b[A\n",
      " 31%|██████████████████████▉                                                    | 11/36 [00:34<01:18,  3.15s/it]\u001b[A\n",
      " 33%|█████████████████████████                                                  | 12/36 [00:37<01:15,  3.15s/it]\u001b[A\n",
      " 36%|███████████████████████████                                                | 13/36 [00:40<01:12,  3.15s/it]\u001b[A\n",
      " 39%|█████████████████████████████▏                                             | 14/36 [00:44<01:09,  3.15s/it]\u001b[A\n",
      " 42%|███████████████████████████████▎                                           | 15/36 [00:47<01:06,  3.15s/it]\u001b[A\n",
      " 44%|█████████████████████████████████▎                                         | 16/36 [00:50<01:03,  3.15s/it]\u001b[A\n",
      " 47%|███████████████████████████████████▍                                       | 17/36 [00:53<00:59,  3.15s/it]\u001b[A\n",
      " 50%|█████████████████████████████████████▌                                     | 18/36 [00:56<00:56,  3.15s/it]\u001b[A\n",
      " 53%|███████████████████████████████████████▌                                   | 19/36 [00:59<00:53,  3.15s/it]\u001b[A\n",
      " 56%|█████████████████████████████████████████▋                                 | 20/36 [01:02<00:50,  3.15s/it]\u001b[A\n",
      " 58%|███████████████████████████████████████████▊                               | 21/36 [01:06<00:47,  3.15s/it]\u001b[A\n",
      " 61%|█████████████████████████████████████████████▊                             | 22/36 [01:09<00:44,  3.15s/it]\u001b[A\n",
      " 64%|███████████████████████████████████████████████▉                           | 23/36 [01:12<00:40,  3.15s/it]\u001b[A\n",
      " 67%|██████████████████████████████████████████████████                         | 24/36 [01:15<00:37,  3.15s/it]\u001b[A\n",
      " 69%|████████████████████████████████████████████████████                       | 25/36 [01:18<00:34,  3.15s/it]\u001b[A\n",
      " 72%|██████████████████████████████████████████████████████▏                    | 26/36 [01:21<00:31,  3.15s/it]\u001b[A\n",
      " 75%|████████████████████████████████████████████████████████▎                  | 27/36 [01:25<00:28,  3.15s/it]\u001b[A\n",
      " 78%|██████████████████████████████████████████████████████████▎                | 28/36 [01:28<00:25,  3.15s/it]\u001b[A\n",
      " 81%|████████████████████████████████████████████████████████████▍              | 29/36 [01:31<00:22,  3.14s/it]\u001b[A\n",
      " 83%|██████████████████████████████████████████████████████████████▌            | 30/36 [01:34<00:18,  3.15s/it]\u001b[A\n",
      " 86%|████████████████████████████████████████████████████████████████▌          | 31/36 [01:37<00:15,  3.15s/it]\u001b[A\n",
      " 89%|██████████████████████████████████████████████████████████████████▋        | 32/36 [01:40<00:12,  3.15s/it]\u001b[A\n",
      " 92%|████████████████████████████████████████████████████████████████████▊      | 33/36 [01:43<00:09,  3.15s/it]\u001b[A\n",
      " 94%|██████████████████████████████████████████████████████████████████████▊    | 34/36 [01:47<00:06,  3.15s/it]\u001b[A\n",
      " 97%|████████████████████████████████████████████████████████████████████████▉  | 35/36 [01:50<00:03,  3.15s/it]\u001b[A\n",
      "100%|███████████████████████████████████████████████████████████████████████████| 36/36 [01:53<00:00,  3.15s/it]\u001b[A\n",
      "100%|████████████████████████████████████████████████████████████████████████████| 3/3 [05:50<00:00, 116.67s/it]\n"
     ]
    }
   ],
   "source": [
    "result_summary = creme.higher_order_interaction_test(model, wt_seq, cre_tiles.copy(),\n",
    "                                                  np.argmin, num_shuffle=5, num_rounds=3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4120e232-5deb-47e4-ad8e-c5a5fce7a15b",
   "metadata": {},
   "source": [
    "Result breakdown"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "04439be1-a7da-4d56-8195-812d45df4e39",
   "metadata": {},
   "source": [
    "During the first iteration, we start with the WT sequence. The initial prediction is the same as WT.  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "bf032d71-d1f9-48b4-96be-bc78966929dc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result_summary[0]['initial_pred'] == wt.mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "102f2105-66ef-43f9-8a93-21297e2767cf",
   "metadata": {},
   "source": [
    "In the first iteration all the tiles are shuffled producing 38 prediction values. The lowest TSS activity value will be selected in this iteration and saved as selected_mean_pred - in this case the point circled in black in panel 1. After this the tile is kept shuffled and iteration 2 starts with the remaining 37 tiles. In each iteration 1 tile is elminated and the plots show the selected tile in each case - the shuffling of which yields the lowest TSS activity."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "5453e25e-5ea7-4631-91f6-217fa6a737df",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of Tiles shuffled in iteration 1: 38\n",
      "Number of Tiles shuffled in iteration 2: 37\n",
      "Number of Tiles shuffled in iteration 3: 36\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1500x400 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(1, len(result_summary.keys()), figsize=[15, 4], sharey=True)\n",
    "shuffled_tiles = []\n",
    "for i in result_summary.keys():\n",
    "    ax = axes[i]\n",
    "    print(f'Number of Tiles shuffled in iteration {i+1}: {result_summary[i][\"preds\"].shape[0]}')\n",
    "    mean_preds = result_summary[i]['preds'].mean(axis=-1).copy()\n",
    "    for s in shuffled_tiles:\n",
    "        mean_preds = np.insert(mean_preds, s, np.nan)\n",
    "    min_i = np.nanargmin(mean_preds)\n",
    "    shuffled_tiles.append(min_i)\n",
    "    \n",
    "    sns.scatterplot(x=range(38), y=mean_preds/wt, color='purple', ax=ax)\n",
    "    ax.set_xlabel('Tile number')\n",
    "    axes[0].set_ylabel('Normalized prediction')\n",
    "    assert np.min(result_summary[i]['preds'].mean(axis=-1)) == result_summary[i]['selected_mean_pred']\n",
    "    sns.scatterplot(x=[min_i], y=[mean_preds[min_i]/wt], s=80, facecolors='none', edgecolors='k', linewidth=2, ax=ax)\n",
    "    ax.annotate(f'selected tile #{min_i}', [min_i+1.5, mean_preds[min_i]/wt]);\n",
    "plt.tight_layout()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "79ed4275-35f5-43d2-958f-1ff172ae3593",
   "metadata": {},
   "source": [
    "Now let's re-create the sequences in each iteration by following the tile order in 'selected tile' to shuffle putative CRE tiles, then obtain predictions and plot those versus the WT."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "ff14f96b-163e-4c18-838b-627d867bc494",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.bin_index = None # switch to saving all track not just TSS bins\n",
    "wt_track = model.predict(wt_seq)[0,:,0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "ca44937c-c56b-4217-9c3f-5f7616ee5316",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "shuffled_tile_seq = wt_seq.copy()\n",
    "per_iteration_preds = []\n",
    "for i in result_summary.keys():\n",
    "    preds = []\n",
    "    tile_start, tile_end = result_summary[i]['selected_tile']\n",
    "    #shuffle this\n",
    "    for j in range(10): # number of shuffles\n",
    "        background_seq = shuffle.dinuc_shuffle(wt_seq)\n",
    "        shuffled_tile_seq[tile_start: tile_end] = background_seq[tile_start: tile_end]\n",
    "        preds.append(model.predict(shuffled_tile_seq)[0,:,0])\n",
    "    per_iteration_preds.append(preds)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "a40848d1-4685-4a7b-8ccd-0ba9f5c36b33",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 2000x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax=utils.plot_track([wt_track], label='WT', color='k', alpha=1)\n",
    "colors = ['purple', 'cyan', 'orange']\n",
    "for i, preds in enumerate(per_iteration_preds):\n",
    "    \n",
    "    utils.plot_track([np.array(preds).mean(axis=0)], zoom=[420, 480], label=f'iteration {i+1}', ax=ax, color=colors[i])\n",
    "ax.set_ylim(0,100)\n",
    "plt.legend();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e3b907e6-ac4a-4684-9c0b-45c8f92c3449",
   "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": "3.9.16"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}