human_cancer_analysis/2_identify_generic_genes_pathways.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Identify generic genes and pathways\n",
    "\n",
    "Studies have found that some genes are more likely to be differentially expressed even across a wide range of experimental designs. These generic genes and subsequent pathways are not necessarily specific to the biological process being studied but instead represent a more systematic change.\n",
    "\n",
    "This notebook identifies generic genes and pathways and then evaluates if those identified are consistent with published findings.\n",
    "\n",
    "**Steps to identify generic genes:**\n",
    "1. Simulates N gene expression experiments using [ponyo](https://github.com/ajlee21/ponyo)\n",
    "2. Perform DE analysis to get association statistics for each gene\n",
    "3. For each gene, aggregate statsitics across all simulated experiments\n",
    "4. Rank genes based on this aggregated statistic (i.e. log fold change, or p-value)\n",
    "\n",
    "\n",
    "**Steps to identify generic gene sets (pathways):**\n",
    "1. Using the same simulated experiments from above, perform GSEA analysis. This analysis will determine whether the genes contained in a gene set are clustered towards the beginning or the end of the ranked list of genes, where genes are ranked by log fold change, indicating a correlation with change in expression.\n",
    "2. For each gene set (pathway), aggregate statistics across all simulated experiments\n",
    "3. Rank gene sets based on this aggregated statistic\n",
    "\n",
    "**Evaluation:**\n",
    "* We want to compare the ranking of genes identified using the above method with the ranking found from [Crow et. al.](https://www.pnas.org/content/pnas/116/13/6491.full.pdf), which identified a set of genes as generic based on how frequently they were found to be DE across 600 experiments\n",
    "* We want to compare the ranking of pathways identified using the above method with the ranking based on the [Powers et. al.](https://www.biorxiv.org/content/10.1101/259440v1.full.pdf) data, where ranking was determined based on the fraction of 442 experiments a pathway was found to be enriched\n",
    "* This comparison will validate our method being used as a way to automatically identify generic genes and pathways."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n",
      "/home/alexandra/anaconda3/envs/generic_expression_new/lib/python3.7/site-packages/matplotlib/__init__.py:886: MatplotlibDeprecationWarning: \n",
      "examples.directory is deprecated; in the future, examples will be found relative to the 'datapath' directory.\n",
      "  \"found relative to the 'datapath' directory.\".format(key))\n"
     ]
    }
   ],
   "source": [
    "%load_ext autoreload\n",
    "%load_ext rpy2.ipython\n",
    "%autoreload 2\n",
    "%matplotlib inline\n",
    "\n",
    "import os\n",
    "import sys\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import pickle\n",
    "import glob\n",
    "import scipy.stats as ss\n",
    "from keras.models import load_model\n",
    "from rpy2.robjects import pandas2ri\n",
    "from ponyo import utils, simulate_expression_data\n",
    "from generic_expression_patterns_modules import process, stats, ranking\n",
    "\n",
    "pandas2ri.activate()\n",
    "\n",
    "np.random.seed(123)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read in config variables\n",
    "base_dir = os.path.abspath(os.path.join(os.getcwd(), \"../\"))\n",
    "\n",
    "config_filename = os.path.abspath(\n",
    "    os.path.join(base_dir, \"configs\", \"config_human_cancer.tsv\")\n",
    ")\n",
    "\n",
    "params = utils.read_config(config_filename)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load params\n",
    "local_dir = params[\"local_dir\"]\n",
    "dataset_name = params[\"dataset_name\"]\n",
    "NN_architecture = params[\"NN_architecture\"]\n",
    "latent_dim = params[\"latent_dim\"]\n",
    "num_runs = params[\"num_simulated\"]\n",
    "project_id = params[\"project_id\"]\n",
    "metadata_col_id = params[\"metadata_colname\"]\n",
    "mapped_template_filename = params[\"mapped_template_filename\"]\n",
    "processed_template_filename = params[\"processed_template_filename\"]\n",
    "normalized_compendium_filename = params[\"normalized_compendium_filename\"]\n",
    "scaler_filename = params[\"scaler_filename\"]\n",
    "col_to_rank_genes = params[\"rank_genes_by\"]\n",
    "col_to_rank_pathways = params[\"rank_pathways_by\"]\n",
    "statistic = params[\"gsea_statistic\"]\n",
    "logFC_name = params[\"DE_logFC_name\"]\n",
    "pvalue_name = params[\"DE_pvalue_name\"]\n",
    "\n",
    "# Load metadata file with grouping assignments for samples\n",
    "sample_id_metadata_filename = os.path.join(\n",
    "    base_dir, dataset_name, \"data\", \"metadata\", f\"{project_id}_process_samples.tsv\"\n",
    ")\n",
    "\n",
    "# Load metadata file with grouping assignments for samples\n",
    "grp_metadata_filename = os.path.join(\n",
    "    base_dir, dataset_name, \"data\", \"metadata\", f\"{project_id}_groups.tsv\"\n",
    ")\n",
    "\n",
    "# Load metadata file with mapping between experiments and associated samples\n",
    "metadata_simulate_filename = os.path.join(\n",
    "    base_dir, dataset_name, \"data\", \"metadata\", \"all_experiments_sample_annotations.csv\"\n",
    ")\n",
    "metadata_delimiter = \",\"\n",
    "experiment_id_colname = \"gse\"\n",
    "\n",
    "# Load pickled file\n",
    "with open(scaler_filename, \"rb\") as scaler_fh:\n",
    "    scaler = pickle.load(scaler_fh)\n",
    "\n",
    "# Percentile threshold to identify generic genes\n",
    "percentile_threshold = 80.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Output files\n",
    "gene_summary_filename = os.path.join(\n",
    "    base_dir, dataset_name, f\"generic_gene_summary_{project_id}.tsv\"\n",
    ")\n",
    "\n",
    "pathway_summary_filename = os.path.join(\n",
    "    base_dir, dataset_name, f\"generic_pathway_summary_{project_id}.tsv\"\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Simulate experiments using selected template experiment\n",
    "\n",
    "Workflow:\n",
    "1. Get the gene expression data for the selected template experiment\n",
    "2. Encode this experiment into a latent space using the trained VAE model\n",
    "3. Linearly shift the encoded template experiment in the latent space\n",
    "4. Decode the samples. This results in a new experiment\n",
    "5. Repeat steps 1-4 to get multiple simulated experiments"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'# Simulate multiple experiments\\n# This step creates the following files in \"<local_dir>/pseudo_experiment/\" directory:\\n#   - selected_simulated_data_SRP012656_<n>.txt\\n#   - selected_simulated_encoded_data_SRP012656_<n>.txt\\n#   - template_normalized_data_SRP012656_test.txt\\n# in which \"<n>\" is an integer in the range of [0, num_runs-1]\\nos.makedirs(os.path.join(local_dir, \"pseudo_experiment\"), exist_ok=True)\\nsimulate_expression_data.shift_template_experiment(\\n    normalized_compendium_filename,\\n    NN_architecture,\\n    latent_dim,\\n    dataset_name,\\n    scaler,\\n    metadata_simulate_filename,\\n    metadata_delimiter,\\n    experiment_id_colname,\\n    metadata_col_id,\\n    project_id,\\n    local_dir,\\n    base_dir,\\n    num_runs,\\n)'"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Simulate multiple experiments\n",
    "# This step creates the following files in \"<local_dir>/pseudo_experiment/\" directory:\n",
    "#   - selected_simulated_data_SRP012656_<n>.txt\n",
    "#   - selected_simulated_encoded_data_SRP012656_<n>.txt\n",
    "#   - template_normalized_data_SRP012656_test.txt\n",
    "# in which \"<n>\" is an integer in the range of [0, num_runs-1]\n",
    "os.makedirs(os.path.join(local_dir, \"pseudo_experiment\"), exist_ok=True)\n",
    "simulate_expression_data.shift_template_experiment(\n",
    "    normalized_compendium_filename,\n",
    "    NN_architecture,\n",
    "    latent_dim,\n",
    "    dataset_name,\n",
    "    scaler,\n",
    "    metadata_simulate_filename,\n",
    "    metadata_delimiter,\n",
    "    experiment_id_colname,\n",
    "    metadata_col_id,\n",
    "    project_id,\n",
    "    local_dir,\n",
    "    base_dir,\n",
    "    num_runs,\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Process template and simulated experiments\n",
    "\n",
    "* Remove samples not required for comparison. Since this experiment contains multiple conditions (i.e. estradiol vs EtOH at 12, 24, and 48 hrs are each considered a different comparison) being tested, we will only include those samples within the same condition.\n",
    "* Make sure ordering of samples matches metadata for proper comparison"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'if not os.path.exists(sample_id_metadata_filename):\\n    sample_id_metadata_filename = None\\n\\nstats.process_samples_for_limma(\\n    mapped_template_filename,\\n    grp_metadata_filename,\\n    processed_template_filename,\\n    sample_id_metadata_filename,\\n)\\n\\nfor i in range(num_runs):\\n    simulated_filename = os.path.join(\\n        local_dir, \"pseudo_experiment\", f\"selected_simulated_data_{project_id}_{i}.txt\"\\n    )\\n    out_simulated_filename = os.path.join(\\n        local_dir,\\n        \"pseudo_experiment\",\\n        f\"selected_simulated_data_{project_id}_{i}_processed.txt\",\\n    )\\n    stats.process_samples_for_limma(\\n        simulated_filename,\\n        grp_metadata_filename,\\n        out_simulated_filename,\\n        sample_id_metadata_filename,\\n    )'"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "if not os.path.exists(sample_id_metadata_filename):\n",
    "    sample_id_metadata_filename = None\n",
    "\n",
    "stats.process_samples_for_limma(\n",
    "    mapped_template_filename,\n",
    "    grp_metadata_filename,\n",
    "    processed_template_filename,\n",
    "    sample_id_metadata_filename,\n",
    ")\n",
    "\n",
    "for i in range(num_runs):\n",
    "    simulated_filename = os.path.join(\n",
    "        local_dir, \"pseudo_experiment\", f\"selected_simulated_data_{project_id}_{i}.txt\"\n",
    "    )\n",
    "    out_simulated_filename = os.path.join(\n",
    "        local_dir,\n",
    "        \"pseudo_experiment\",\n",
    "        f\"selected_simulated_data_{project_id}_{i}_processed.txt\",\n",
    "    )\n",
    "    stats.process_samples_for_limma(\n",
    "        simulated_filename,\n",
    "        grp_metadata_filename,\n",
    "        out_simulated_filename,\n",
    "        sample_id_metadata_filename,\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Differential expression analysis\n",
    "\n",
    "The gene expression dataset is array-based so we will use Limma in this case"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create subdirectory: \"<local_dir>/DE_stats/\"\n",
    "os.makedirs(os.path.join(local_dir, \"DE_stats\"), exist_ok=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "R[write to console]: Loading required package: S4Vectors\n",
      "\n",
      "R[write to console]: Loading required package: stats4\n",
      "\n",
      "R[write to console]: Loading required package: BiocGenerics\n",
      "\n",
      "R[write to console]: Loading required package: parallel\n",
      "\n",
      "R[write to console]: \n",
      "Attaching package: ‘BiocGenerics’\n",
      "\n",
      "\n",
      "R[write to console]: The following objects are masked from ‘package:parallel’:\n",
      "\n",
      "    clusterApply, clusterApplyLB, clusterCall, clusterEvalQ,\n",
      "    clusterExport, clusterMap, parApply, parCapply, parLapply,\n",
      "    parLapplyLB, parRapply, parSapply, parSapplyLB\n",
      "\n",
      "\n",
      "R[write to console]: The following object is masked from ‘package:limma’:\n",
      "\n",
      "    plotMA\n",
      "\n",
      "\n",
      "R[write to console]: The following objects are masked from ‘package:stats’:\n",
      "\n",
      "    IQR, mad, sd, var, xtabs\n",
      "\n",
      "\n",
      "R[write to console]: The following objects are masked from ‘package:base’:\n",
      "\n",
      "    anyDuplicated, append, as.data.frame, basename, cbind, colnames,\n",
      "    dirname, do.call, duplicated, eval, evalq, Filter, Find, get, grep,\n",
      "    grepl, intersect, is.unsorted, lapply, Map, mapply, match, mget,\n",
      "    order, paste, pmax, pmax.int, pmin, pmin.int, Position, rank,\n",
      "    rbind, Reduce, rownames, sapply, setdiff, sort, table, tapply,\n",
      "    union, unique, unsplit, which, which.max, which.min\n",
      "\n",
      "\n",
      "R[write to console]: \n",
      "Attaching package: ‘S4Vectors’\n",
      "\n",
      "\n",
      "R[write to console]: The following object is masked from ‘package:base’:\n",
      "\n",
      "    expand.grid\n",
      "\n",
      "\n",
      "R[write to console]: Loading required package: IRanges\n",
      "\n",
      "R[write to console]: Loading required package: GenomicRanges\n",
      "\n",
      "R[write to console]: Loading required package: GenomeInfoDb\n",
      "\n",
      "R[write to console]: Loading required package: SummarizedExperiment\n",
      "\n",
      "R[write to console]: Loading required package: Biobase\n",
      "\n",
      "R[write to console]: Welcome to Bioconductor\n",
      "\n",
      "    Vignettes contain introductory material; view with\n",
      "    'browseVignettes()'. To cite Bioconductor, see\n",
      "    'citation(\"Biobase\")', and for packages 'citation(\"pkgname\")'.\n",
      "\n",
      "\n",
      "R[write to console]: Loading required package: DelayedArray\n",
      "\n",
      "R[write to console]: Loading required package: matrixStats\n",
      "\n",
      "R[write to console]: \n",
      "Attaching package: ‘matrixStats’\n",
      "\n",
      "\n",
      "R[write to console]: The following objects are masked from ‘package:Biobase’:\n",
      "\n",
      "    anyMissing, rowMedians\n",
      "\n",
      "\n",
      "R[write to console]: Loading required package: BiocParallel\n",
      "\n",
      "R[write to console]: \n",
      "Attaching package: ‘DelayedArray’\n",
      "\n",
      "\n",
      "R[write to console]: The following objects are masked from ‘package:matrixStats’:\n",
      "\n",
      "    colMaxs, colMins, colRanges, rowMaxs, rowMins, rowRanges\n",
      "\n",
      "\n",
      "R[write to console]: The following objects are masked from ‘package:base’:\n",
      "\n",
      "    aperm, apply, rowsum\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1] 31\n"
     ]
    }
   ],
   "source": [
    "%%R -i grp_metadata_filename -i project_id -i processed_template_filename -i local_dir -i base_dir\n",
    "\n",
    "source(paste0(base_dir, '/generic_expression_patterns_modules/DE_analysis.R'))\n",
    "\n",
    "# File created: \"<local_dir>/DE_stats/DE_stats_template_data_SRP012656_real.txt\"\n",
    "get_DE_stats_limma(grp_metadata_filename,\n",
    "                   project_id,\n",
    "                   processed_template_filename,\n",
    "                   \"template\",\n",
    "                   local_dir,\n",
    "                   \"real\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(33, 6)\n"
     ]
    }
   ],
   "source": [
    "# Check number of DEGs\n",
    "template_DE_stats_filename = os.path.join(\n",
    "    local_dir, \"DE_stats\", f\"DE_stats_template_data_{project_id}_real.txt\"\n",
    ")\n",
    "\n",
    "template_DE_stats = pd.read_csv(\n",
    "    template_DE_stats_filename, sep=\"\\t\", header=0, index_col=0\n",
    ")\n",
    "\n",
    "selected = template_DE_stats[\n",
    "    (template_DE_stats[\"adj.P.Val\"] < 0.05) & (abs(template_DE_stats[\"logFC\"]) > 1)\n",
    "]\n",
    "print(selected.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "%%R -i grp_metadata_filename -i project_id -i base_dir -i local_dir -i num_runs\n",
    "\n",
    "source(paste0(base_dir, '/generic_expression_patterns_modules/DE_analysis.R'))\n",
    "\n",
    "# Files created: \"<local_dir>/DE_stats/DE_stats_simulated_data_SRP012656_<n>.txt\"\n",
    "for (i in 0:(num_runs-1)){\n",
    "    simulated_data_filename <- paste(local_dir,\n",
    "                                     \"pseudo_experiment/selected_simulated_data_\",\n",
    "                                     project_id,\n",
    "                                     \"_\",\n",
    "                                     i,\n",
    "                                     \"_processed.txt\",\n",
    "                                     sep = \"\")\n",
    "\n",
    "    get_DE_stats_limma(grp_metadata_filename,\n",
    "                       project_id,\n",
    "                       simulated_data_filename,\n",
    "                       \"simulated\",\n",
    "                       local_dir,\n",
    "                       i)\n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Rank genes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/alexandra/anaconda3/envs/generic_expression_new/lib/python3.7/site-packages/pandas/core/reshape/merge.py:544: UserWarning: merging between different levels can give an unintended result (2 levels on the left, 1 on the right)\n",
      "  warnings.warn(msg, UserWarning)\n"
     ]
    }
   ],
   "source": [
    "analysis_type = \"DE\"\n",
    "template_DE_stats, simulated_DE_summary_stats = ranking.process_and_rank_genes_pathways(\n",
    "    template_DE_stats_filename,\n",
    "    local_dir,\n",
    "    num_runs,\n",
    "    project_id,\n",
    "    analysis_type,\n",
    "    col_to_rank_genes,\n",
    "    logFC_name,\n",
    "    pvalue_name,\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Gene summary table"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Gene ID</th>\n",
       "      <th>Adj P-value (Real)</th>\n",
       "      <th>Rank (Real)</th>\n",
       "      <th>abs(logFC) (Real)</th>\n",
       "      <th>logFC (Real)</th>\n",
       "      <th>Median adj p-value (simulated)</th>\n",
       "      <th>Rank (simulated)</th>\n",
       "      <th>Percentile (simulated)</th>\n",
       "      <th>Percent DE (simulated)</th>\n",
       "      <th>Mean abs(logFC) (simulated)</th>\n",
       "      <th>Std deviation (simulated)</th>\n",
       "      <th>Number of experiments (simulated)</th>\n",
       "      <th>Z score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>EGR3</th>\n",
       "      <td>EGR3</td>\n",
       "      <td>4.265860e-09</td>\n",
       "      <td>6763.0</td>\n",
       "      <td>3.945807</td>\n",
       "      <td>3.945807</td>\n",
       "      <td>0.334186</td>\n",
       "      <td>5489.5</td>\n",
       "      <td>81.166815</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.056253</td>\n",
       "      <td>0.057537</td>\n",
       "      <td>25</td>\n",
       "      <td>67.601429</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>PGR</th>\n",
       "      <td>PGR</td>\n",
       "      <td>9.812526e-07</td>\n",
       "      <td>6762.0</td>\n",
       "      <td>2.620937</td>\n",
       "      <td>2.620937</td>\n",
       "      <td>0.719110</td>\n",
       "      <td>3062.5</td>\n",
       "      <td>45.275067</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.027507</td>\n",
       "      <td>0.025951</td>\n",
       "      <td>25</td>\n",
       "      <td>99.936991</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>PRSS23</th>\n",
       "      <td>PRSS23</td>\n",
       "      <td>1.683715e-06</td>\n",
       "      <td>6761.0</td>\n",
       "      <td>2.557864</td>\n",
       "      <td>2.557864</td>\n",
       "      <td>0.554417</td>\n",
       "      <td>5568.5</td>\n",
       "      <td>82.335108</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.053680</td>\n",
       "      <td>0.045390</td>\n",
       "      <td>25</td>\n",
       "      <td>55.170351</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>MYBL1</th>\n",
       "      <td>MYBL1</td>\n",
       "      <td>3.661078e-06</td>\n",
       "      <td>6760.0</td>\n",
       "      <td>1.919970</td>\n",
       "      <td>1.919970</td>\n",
       "      <td>0.148952</td>\n",
       "      <td>6748.0</td>\n",
       "      <td>99.778172</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.127000</td>\n",
       "      <td>0.063397</td>\n",
       "      <td>25</td>\n",
       "      <td>28.281728</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CXCL12</th>\n",
       "      <td>CXCL12</td>\n",
       "      <td>1.083478e-05</td>\n",
       "      <td>6759.0</td>\n",
       "      <td>1.834330</td>\n",
       "      <td>1.834330</td>\n",
       "      <td>0.706058</td>\n",
       "      <td>5209.0</td>\n",
       "      <td>77.018634</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.040893</td>\n",
       "      <td>0.038614</td>\n",
       "      <td>25</td>\n",
       "      <td>46.445058</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Gene ID  Adj P-value (Real)  Rank (Real)  abs(logFC) (Real)  \\\n",
       "EGR3      EGR3        4.265860e-09       6763.0           3.945807   \n",
       "PGR        PGR        9.812526e-07       6762.0           2.620937   \n",
       "PRSS23  PRSS23        1.683715e-06       6761.0           2.557864   \n",
       "MYBL1    MYBL1        3.661078e-06       6760.0           1.919970   \n",
       "CXCL12  CXCL12        1.083478e-05       6759.0           1.834330   \n",
       "\n",
       "        logFC (Real)  Median adj p-value (simulated)  Rank (simulated)  \\\n",
       "EGR3        3.945807                        0.334186            5489.5   \n",
       "PGR         2.620937                        0.719110            3062.5   \n",
       "PRSS23      2.557864                        0.554417            5568.5   \n",
       "MYBL1       1.919970                        0.148952            6748.0   \n",
       "CXCL12      1.834330                        0.706058            5209.0   \n",
       "\n",
       "        Percentile (simulated)  Percent DE (simulated)  \\\n",
       "EGR3                 81.166815                     0.0   \n",
       "PGR                  45.275067                     0.0   \n",
       "PRSS23               82.335108                     0.0   \n",
       "MYBL1                99.778172                     0.0   \n",
       "CXCL12               77.018634                     0.0   \n",
       "\n",
       "        Mean abs(logFC) (simulated)  Std deviation (simulated)  \\\n",
       "EGR3                       0.056253                   0.057537   \n",
       "PGR                        0.027507                   0.025951   \n",
       "PRSS23                     0.053680                   0.045390   \n",
       "MYBL1                      0.127000                   0.063397   \n",
       "CXCL12                     0.040893                   0.038614   \n",
       "\n",
       "        Number of experiments (simulated)    Z score  \n",
       "EGR3                                   25  67.601429  \n",
       "PGR                                    25  99.936991  \n",
       "PRSS23                                 25  55.170351  \n",
       "MYBL1                                  25  28.281728  \n",
       "CXCL12                                 25  46.445058  "
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "summary_gene_ranks = ranking.generate_summary_table(\n",
    "    template_DE_stats_filename,\n",
    "    template_DE_stats,\n",
    "    simulated_DE_summary_stats,\n",
    "    col_to_rank_genes,\n",
    "    local_dir,\n",
    "    \"gene\",\n",
    "    params,\n",
    ")\n",
    "\n",
    "summary_gene_ranks.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Gene ID                              False\n",
       "Adj P-value (Real)                   False\n",
       "Rank (Real)                          False\n",
       "abs(logFC) (Real)                    False\n",
       "logFC (Real)                         False\n",
       "Median adj p-value (simulated)       False\n",
       "Rank (simulated)                     False\n",
       "Percentile (simulated)               False\n",
       "Percent DE (simulated)               False\n",
       "Mean abs(logFC) (simulated)          False\n",
       "Std deviation (simulated)            False\n",
       "Number of experiments (simulated)    False\n",
       "Z score                              False\n",
       "dtype: bool"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check if there is an NaN values, there should not be\n",
    "summary_gene_ranks.isna().any()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create `gene_summary_fielname`\n",
    "summary_gene_ranks.to_csv(gene_summary_filename, sep=\"\\t\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Compare gene ranking\n",
    "Studies have found that some genes are more likely to be differentially expressed even across a wide range of experimental designs. These *generic genes* are not necessarily specific to the biological process being studied but instead represent a more systematic change.\n",
    "\n",
    "We want to compare the ability to detect these generic genes using our method vs those found by [Crow et. al. publication](https://www.pnas.org/content/pnas/116/13/6491.full.pdf). Their genes are ranked 0 = not commonly DE; 1 = commonly DE. Genes were ranked by the number differentially expressed gene sets a gene appeared in across 600 experiments."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "r = 0.25522576510350836\n",
      "p = 7.284866435208007e-96\n",
      "ci_low = 0.23206295322579806\n",
      "ci_high = 0.27864926060564976\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/alexandra/anaconda3/envs/generic_expression_new/lib/python3.7/site-packages/matplotlib/font_manager.py:1238: UserWarning: findfont: Font family ['Verdana'] not found. Falling back to DejaVu Sans.\n",
      "  (prop.get_family(), self.defaultFamily[fontext]))\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x432 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Get generic genes identified by Crow et. al.\n",
    "DE_prior_filename = params[\"reference_gene_filename\"]\n",
    "ref_gene_col = params[\"reference_gene_name_col\"]\n",
    "ref_rank_col = params[\"reference_rank_col\"]\n",
    "\n",
    "figure_filename = f\"gene_ranking_{col_to_rank_genes}.svg\"\n",
    "\n",
    "corr, shared_ranking = ranking.compare_gene_ranking(\n",
    "    summary_gene_ranks, DE_prior_filename, ref_gene_col, ref_rank_col, figure_filename\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Hypergeometric test:\n",
    "\n",
    "Given $N$ number of genes with $K$ common genes in Crow et al. SOPHIE identifies $n$ genes as being common. What is the probability that $k$ of the genes identified by SOPHIE are also common in Crow et al.? What is the probability of drawing $k$ or more concordant genes?\n",
    "\n",
    "This was a way for us to quantify the correlation between SOPHIE and Crow et al common findings, since the correlation coefficient wasn't very convincing since we're considering all genes in addition to the common ones"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "num_Crow_genes = shared_ranking.shape[0]\n",
    "num_generic_Crow_genes = shared_ranking.query(f\"{ref_rank_col}>=80.0\").shape[0]\n",
    "num_generic_SOPHIE_genes = shared_ranking[\n",
    "    shared_ranking[\"Percentile (simulated)\"] >= percentile_threshold\n",
    "].shape[0]\n",
    "num_concordant_generic_genes = shared_ranking[\n",
    "    (shared_ranking[ref_rank_col] >= percentile_threshold)\n",
    "    & (shared_ranking[\"Percentile (simulated)\"] >= percentile_threshold)\n",
    "].shape[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6410\n",
      "1441\n",
      "1331\n",
      "374\n"
     ]
    }
   ],
   "source": [
    "print(num_Crow_genes)\n",
    "print(num_generic_Crow_genes)\n",
    "print(num_generic_SOPHIE_genes)\n",
    "print(num_concordant_generic_genes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.514841880375263e-08\n"
     ]
    }
   ],
   "source": [
    "p = ss.hypergeom.sf(\n",
    "    num_concordant_generic_genes,\n",
    "    num_Crow_genes,\n",
    "    num_generic_Crow_genes,\n",
    "    num_generic_SOPHIE_genes,\n",
    ")\n",
    "print(p)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Takeaway:**\n",
    "* Here we are comparing gene percentiles obtained from a SOPHIE trained on Powers et al. vs gene percentiles obtained from manual curation using Crow et al. The plot shows that there is a high correlation between those very high and low ranked genes -- high correlation at the extremes but there is a lot of noise in the middle.\n",
    "* While the two datasets used the same array platform to generate data, the datasets have different compositions – Crow et al. is a heterogenous mixture of different types of experiments while Power et al. is specifically cancer cell lines treated with small molecules. The consistency we observe in the common DEGs despite the differences in context demonstrates that many common DEGs are differentially expressed regardless of the context."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### GSEA\n",
    "**Goal:** To detect modest but coordinated changes in prespecified sets of related genes (i.e. those genes in the same pathway or share the same GO term).\n",
    "\n",
    "1. Rank all genes using DE association statistics.\n",
    "2. An enrichment score (ES) is defined as the maximum distance from the middle of the ranked list. Thus, the enrichment score indicates whether the genes contained in a gene set are clustered towards the beginning or the end of the ranked list (indicating a correlation with change in expression).\n",
    "3. Estimate the statistical significance of the ES by a phenotypic-based permutation test in order to produce a null distribution for the ES (i.e. scores based on permuted phenotype)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create \"<local_dir>/GSEA_stats/\" subdirectory\n",
    "os.makedirs(os.path.join(local_dir, \"GSA_stats\"), exist_ok=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load pathway data\n",
    "hallmark_DB_filename = params[\"pathway_DB_filename\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "R[write to console]: Loading required package: Rcpp\n",
      "\n"
     ]
    }
   ],
   "source": [
    "%%R -i base_dir -i template_DE_stats_filename -i hallmark_DB_filename -i statistic -i local_dir -o template_enriched_pathways\n",
    "\n",
    "source(paste0(base_dir, '/generic_expression_patterns_modules/GSEA_analysis.R'))\n",
    "\n",
    "out_filename <- paste(local_dir,\n",
    "                     \"GSA_stats/GSEA_stats_template_data_\",\n",
    "                     project_id,\n",
    "                     \"_real.txt\",\n",
    "                     sep = \"\")\n",
    "\n",
    "template_enriched_pathways <- find_enriched_pathways(template_DE_stats_filename, hallmark_DB_filename, statistic)\n",
    "template_enriched_pathways <- as.data.frame(template_enriched_pathways[1:7])\n",
    "\n",
    "write.table(template_enriched_pathways, file = out_filename, row.names = F, sep = \"\\t\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(50, 7)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>pathway</th>\n",
       "      <th>pval</th>\n",
       "      <th>padj</th>\n",
       "      <th>ES</th>\n",
       "      <th>NES</th>\n",
       "      <th>nMoreExtreme</th>\n",
       "      <th>size</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>HALLMARK_ESTROGEN_RESPONSE_EARLY</td>\n",
       "      <td>0.000173</td>\n",
       "      <td>0.002917</td>\n",
       "      <td>0.854415</td>\n",
       "      <td>2.961952</td>\n",
       "      <td>0.0</td>\n",
       "      <td>65</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>HALLMARK_ESTROGEN_RESPONSE_LATE</td>\n",
       "      <td>0.000174</td>\n",
       "      <td>0.002917</td>\n",
       "      <td>0.826175</td>\n",
       "      <td>2.791508</td>\n",
       "      <td>0.0</td>\n",
       "      <td>57</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>HALLMARK_INTERFERON_GAMMA_RESPONSE</td>\n",
       "      <td>0.000233</td>\n",
       "      <td>0.002917</td>\n",
       "      <td>-0.557724</td>\n",
       "      <td>-1.960252</td>\n",
       "      <td>0.0</td>\n",
       "      <td>53</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>HALLMARK_MYC_TARGETS_V1</td>\n",
       "      <td>0.000175</td>\n",
       "      <td>0.002917</td>\n",
       "      <td>0.750681</td>\n",
       "      <td>2.495992</td>\n",
       "      <td>0.0</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>HALLMARK_E2F_TARGETS</td>\n",
       "      <td>0.000719</td>\n",
       "      <td>0.007189</td>\n",
       "      <td>0.625182</td>\n",
       "      <td>1.964366</td>\n",
       "      <td>3.0</td>\n",
       "      <td>39</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>HALLMARK_MYC_TARGETS_V2</td>\n",
       "      <td>0.001129</td>\n",
       "      <td>0.009407</td>\n",
       "      <td>0.804683</td>\n",
       "      <td>1.980572</td>\n",
       "      <td>5.0</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>HALLMARK_INTERFERON_ALPHA_RESPONSE</td>\n",
       "      <td>0.002863</td>\n",
       "      <td>0.019120</td>\n",
       "      <td>-0.627691</td>\n",
       "      <td>-1.840370</td>\n",
       "      <td>12.0</td>\n",
       "      <td>24</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>46</th>\n",
       "      <td>HALLMARK_UNFOLDED_PROTEIN_RESPONSE</td>\n",
       "      <td>0.003059</td>\n",
       "      <td>0.019120</td>\n",
       "      <td>0.589271</td>\n",
       "      <td>1.840449</td>\n",
       "      <td>16.0</td>\n",
       "      <td>38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>HALLMARK_INFLAMMATORY_RESPONSE</td>\n",
       "      <td>0.005988</td>\n",
       "      <td>0.033267</td>\n",
       "      <td>-0.405263</td>\n",
       "      <td>-1.551020</td>\n",
       "      <td>24.0</td>\n",
       "      <td>81</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>HALLMARK_APICAL_SURFACE</td>\n",
       "      <td>0.007117</td>\n",
       "      <td>0.035587</td>\n",
       "      <td>-0.729707</td>\n",
       "      <td>-1.767291</td>\n",
       "      <td>33.0</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                               pathway      pval      padj        ES  \\\n",
       "15    HALLMARK_ESTROGEN_RESPONSE_EARLY  0.000173  0.002917  0.854415   \n",
       "16     HALLMARK_ESTROGEN_RESPONSE_LATE  0.000174  0.002917  0.826175   \n",
       "27  HALLMARK_INTERFERON_GAMMA_RESPONSE  0.000233  0.002917 -0.557724   \n",
       "32             HALLMARK_MYC_TARGETS_V1  0.000175  0.002917  0.750681   \n",
       "13                HALLMARK_E2F_TARGETS  0.000719  0.007189  0.625182   \n",
       "33             HALLMARK_MYC_TARGETS_V2  0.001129  0.009407  0.804683   \n",
       "26  HALLMARK_INTERFERON_ALPHA_RESPONSE  0.002863  0.019120 -0.627691   \n",
       "46  HALLMARK_UNFOLDED_PROTEIN_RESPONSE  0.003059  0.019120  0.589271   \n",
       "25      HALLMARK_INFLAMMATORY_RESPONSE  0.005988  0.033267 -0.405263   \n",
       "6              HALLMARK_APICAL_SURFACE  0.007117  0.035587 -0.729707   \n",
       "\n",
       "         NES  nMoreExtreme  size  \n",
       "15  2.961952           0.0    65  \n",
       "16  2.791508           0.0    57  \n",
       "27 -1.960252           0.0    53  \n",
       "32  2.495992           0.0    52  \n",
       "13  1.964366           3.0    39  \n",
       "33  1.980572           5.0    13  \n",
       "26 -1.840370          12.0    24  \n",
       "46  1.840449          16.0    38  \n",
       "25 -1.551020          24.0    81  \n",
       "6  -1.767291          33.0    11  "
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(template_enriched_pathways.shape)\n",
    "template_enriched_pathways[template_enriched_pathways[\"padj\"] < 0.05].sort_values(\n",
    "    by=\"padj\"\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Quick check:** Looks like enriched pathways are consistent with estradiol being estrogen hormone treatment."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "%%R -i project_id -i local_dir -i hallmark_DB_filename -i num_runs -i statistic -i base_dir\n",
    "\n",
    "source(paste0(base_dir, '/generic_expression_patterns_modules/GSEA_analysis.R'))\n",
    "\n",
    "# New files created: \"<local_dir>/GSEA_stats/GSEA_stats_simulated_data_<project_id>_<n>.txt\"\n",
    "for (i in 0:(num_runs-1)) {\n",
    "    simulated_DE_stats_filename <- paste(local_dir,\n",
    "                                     \"DE_stats/DE_stats_simulated_data_\",\n",
    "                                     project_id,\n",
    "                                     \"_\",\n",
    "                                     i,\n",
    "                                     \".txt\",\n",
    "                                     sep = \"\")\n",
    "\n",
    "    out_filename <- paste(local_dir,\n",
    "                     \"GSA_stats/GSEA_stats_simulated_data_\",\n",
    "                     project_id,\n",
    "                     \"_\",\n",
    "                     i,\n",
    "                     \".txt\",\n",
    "                     sep = \"\")\n",
    "\n",
    "    enriched_pathways <- find_enriched_pathways(simulated_DE_stats_filename, hallmark_DB_filename, statistic)\n",
    "\n",
    "    write.table(as.data.frame(enriched_pathways[1:7]), file = out_filename, row.names = F, sep = \"\\t\")\n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Rank pathways"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "analysis_type = \"GSA\"\n",
    "template_GSEA_stats_filename = os.path.join(\n",
    "    local_dir, \"GSA_stats\", f\"GSEA_stats_template_data_{project_id}_real.txt\"\n",
    ")\n",
    "(\n",
    "    template_GSEA_stats,\n",
    "    simulated_GSEA_summary_stats,\n",
    ") = ranking.process_and_rank_genes_pathways(\n",
    "    template_GSEA_stats_filename,\n",
    "    local_dir,\n",
    "    num_runs,\n",
    "    project_id,\n",
    "    analysis_type,\n",
    "    col_to_rank_pathways,\n",
    "    logFC_name,\n",
    "    pvalue_name,\n",
    "    \"GSEA\",\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Pathway summary table"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/alexandra/anaconda3/envs/generic_expression_new/lib/python3.7/site-packages/pandas/core/reshape/merge.py:544: UserWarning: merging between different levels can give an unintended result (1 levels on the left, 2 on the right)\n",
      "  warnings.warn(msg, UserWarning)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Pathway ID</th>\n",
       "      <th>Adj P-value (Real)</th>\n",
       "      <th>Rank (Real)</th>\n",
       "      <th>padj (Real)</th>\n",
       "      <th>Median adj p-value (simulated)</th>\n",
       "      <th>Rank (simulated)</th>\n",
       "      <th>Percentile (simulated)</th>\n",
       "      <th>Mean padj (simulated)</th>\n",
       "      <th>Std deviation (simulated)</th>\n",
       "      <th>Number of experiments (simulated)</th>\n",
       "      <th>Z score</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pathway</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>HALLMARK_G2M_CHECKPOINT</th>\n",
       "      <td>HALLMARK_G2M_CHECKPOINT</td>\n",
       "      <td>0.057869</td>\n",
       "      <td>40.0</td>\n",
       "      <td>0.057869</td>\n",
       "      <td>0.001688</td>\n",
       "      <td>49.0</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>0.010225</td>\n",
       "      <td>0.017949</td>\n",
       "      <td>25</td>\n",
       "      <td>2.654366</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>HALLMARK_MYC_TARGETS_V1</th>\n",
       "      <td>HALLMARK_MYC_TARGETS_V1</td>\n",
       "      <td>0.002917</td>\n",
       "      <td>48.5</td>\n",
       "      <td>0.002917</td>\n",
       "      <td>0.001688</td>\n",
       "      <td>49.0</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>0.011698</td>\n",
       "      <td>0.024162</td>\n",
       "      <td>25</td>\n",
       "      <td>-0.363438</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>HALLMARK_E2F_TARGETS</th>\n",
       "      <td>HALLMARK_E2F_TARGETS</td>\n",
       "      <td>0.007189</td>\n",
       "      <td>46.0</td>\n",
       "      <td>0.007189</td>\n",
       "      <td>0.001688</td>\n",
       "      <td>49.0</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>0.007393</td>\n",
       "      <td>0.013072</td>\n",
       "      <td>25</td>\n",
       "      <td>-0.015574</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>HALLMARK_MYC_TARGETS_V2</th>\n",
       "      <td>HALLMARK_MYC_TARGETS_V2</td>\n",
       "      <td>0.009407</td>\n",
       "      <td>45.0</td>\n",
       "      <td>0.009407</td>\n",
       "      <td>0.010768</td>\n",
       "      <td>47.0</td>\n",
       "      <td>95.833333</td>\n",
       "      <td>0.071022</td>\n",
       "      <td>0.205818</td>\n",
       "      <td>25</td>\n",
       "      <td>-0.299367</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>HALLMARK_MTORC1_SIGNALING</th>\n",
       "      <td>HALLMARK_MTORC1_SIGNALING</td>\n",
       "      <td>0.076419</td>\n",
       "      <td>39.0</td>\n",
       "      <td>0.076419</td>\n",
       "      <td>0.012651</td>\n",
       "      <td>46.0</td>\n",
       "      <td>93.750000</td>\n",
       "      <td>0.135693</td>\n",
       "      <td>0.269983</td>\n",
       "      <td>25</td>\n",
       "      <td>-0.219549</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>HALLMARK_DNA_REPAIR</th>\n",
       "      <td>HALLMARK_DNA_REPAIR</td>\n",
       "      <td>0.304723</td>\n",
       "      <td>31.0</td>\n",
       "      <td>0.304723</td>\n",
       "      <td>0.012845</td>\n",
       "      <td>45.0</td>\n",
       "      <td>91.666667</td>\n",
       "      <td>0.159171</td>\n",
       "      <td>0.309746</td>\n",
       "      <td>25</td>\n",
       "      <td>0.469907</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>HALLMARK_OXIDATIVE_PHOSPHORYLATION</th>\n",
       "      <td>HALLMARK_OXIDATIVE_PHOSPHORYLATION</td>\n",
       "      <td>0.303844</td>\n",
       "      <td>33.0</td>\n",
       "      <td>0.303844</td>\n",
       "      <td>0.018769</td>\n",
       "      <td>44.0</td>\n",
       "      <td>89.583333</td>\n",
       "      <td>0.248496</td>\n",
       "      <td>0.407086</td>\n",
       "      <td>25</td>\n",
       "      <td>0.135960</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>HALLMARK_INFLAMMATORY_RESPONSE</th>\n",
       "      <td>HALLMARK_INFLAMMATORY_RESPONSE</td>\n",
       "      <td>0.033267</td>\n",
       "      <td>42.0</td>\n",
       "      <td>0.033267</td>\n",
       "      <td>0.049409</td>\n",
       "      <td>43.0</td>\n",
       "      <td>87.500000</td>\n",
       "      <td>0.242602</td>\n",
       "      <td>0.369390</td>\n",
       "      <td>22</td>\n",
       "      <td>-0.566707</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>HALLMARK_ALLOGRAFT_REJECTION</th>\n",
       "      <td>HALLMARK_ALLOGRAFT_REJECTION</td>\n",
       "      <td>0.501264</td>\n",
       "      <td>18.0</td>\n",
       "      <td>0.501264</td>\n",
       "      <td>0.053930</td>\n",
       "      <td>42.0</td>\n",
       "      <td>85.416667</td>\n",
       "      <td>0.224399</td>\n",
       "      <td>0.349003</td>\n",
       "      <td>25</td>\n",
       "      <td>0.793303</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>HALLMARK_UNFOLDED_PROTEIN_RESPONSE</th>\n",
       "      <td>HALLMARK_UNFOLDED_PROTEIN_RESPONSE</td>\n",
       "      <td>0.019120</td>\n",
       "      <td>43.5</td>\n",
       "      <td>0.019120</td>\n",
       "      <td>0.054696</td>\n",
       "      <td>41.0</td>\n",
       "      <td>83.333333</td>\n",
       "      <td>0.166565</td>\n",
       "      <td>0.266656</td>\n",
       "      <td>25</td>\n",
       "      <td>-0.552943</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                            Pathway ID  \\\n",
       "pathway                                                                  \n",
       "HALLMARK_G2M_CHECKPOINT                        HALLMARK_G2M_CHECKPOINT   \n",
       "HALLMARK_MYC_TARGETS_V1                        HALLMARK_MYC_TARGETS_V1   \n",
       "HALLMARK_E2F_TARGETS                              HALLMARK_E2F_TARGETS   \n",
       "HALLMARK_MYC_TARGETS_V2                        HALLMARK_MYC_TARGETS_V2   \n",
       "HALLMARK_MTORC1_SIGNALING                    HALLMARK_MTORC1_SIGNALING   \n",
       "HALLMARK_DNA_REPAIR                                HALLMARK_DNA_REPAIR   \n",
       "HALLMARK_OXIDATIVE_PHOSPHORYLATION  HALLMARK_OXIDATIVE_PHOSPHORYLATION   \n",
       "HALLMARK_INFLAMMATORY_RESPONSE          HALLMARK_INFLAMMATORY_RESPONSE   \n",
       "HALLMARK_ALLOGRAFT_REJECTION              HALLMARK_ALLOGRAFT_REJECTION   \n",
       "HALLMARK_UNFOLDED_PROTEIN_RESPONSE  HALLMARK_UNFOLDED_PROTEIN_RESPONSE   \n",
       "\n",
       "                                    Adj P-value (Real)  Rank (Real)  \\\n",
       "pathway                                                               \n",
       "HALLMARK_G2M_CHECKPOINT                       0.057869         40.0   \n",
       "HALLMARK_MYC_TARGETS_V1                       0.002917         48.5   \n",
       "HALLMARK_E2F_TARGETS                          0.007189         46.0   \n",
       "HALLMARK_MYC_TARGETS_V2                       0.009407         45.0   \n",
       "HALLMARK_MTORC1_SIGNALING                     0.076419         39.0   \n",
       "HALLMARK_DNA_REPAIR                           0.304723         31.0   \n",
       "HALLMARK_OXIDATIVE_PHOSPHORYLATION            0.303844         33.0   \n",
       "HALLMARK_INFLAMMATORY_RESPONSE                0.033267         42.0   \n",
       "HALLMARK_ALLOGRAFT_REJECTION                  0.501264         18.0   \n",
       "HALLMARK_UNFOLDED_PROTEIN_RESPONSE            0.019120         43.5   \n",
       "\n",
       "                                    padj (Real)  \\\n",
       "pathway                                           \n",
       "HALLMARK_G2M_CHECKPOINT                0.057869   \n",
       "HALLMARK_MYC_TARGETS_V1                0.002917   \n",
       "HALLMARK_E2F_TARGETS                   0.007189   \n",
       "HALLMARK_MYC_TARGETS_V2                0.009407   \n",
       "HALLMARK_MTORC1_SIGNALING              0.076419   \n",
       "HALLMARK_DNA_REPAIR                    0.304723   \n",
       "HALLMARK_OXIDATIVE_PHOSPHORYLATION     0.303844   \n",
       "HALLMARK_INFLAMMATORY_RESPONSE         0.033267   \n",
       "HALLMARK_ALLOGRAFT_REJECTION           0.501264   \n",
       "HALLMARK_UNFOLDED_PROTEIN_RESPONSE     0.019120   \n",
       "\n",
       "                                    Median adj p-value (simulated)  \\\n",
       "pathway                                                              \n",
       "HALLMARK_G2M_CHECKPOINT                                   0.001688   \n",
       "HALLMARK_MYC_TARGETS_V1                                   0.001688   \n",
       "HALLMARK_E2F_TARGETS                                      0.001688   \n",
       "HALLMARK_MYC_TARGETS_V2                                   0.010768   \n",
       "HALLMARK_MTORC1_SIGNALING                                 0.012651   \n",
       "HALLMARK_DNA_REPAIR                                       0.012845   \n",
       "HALLMARK_OXIDATIVE_PHOSPHORYLATION                        0.018769   \n",
       "HALLMARK_INFLAMMATORY_RESPONSE                            0.049409   \n",
       "HALLMARK_ALLOGRAFT_REJECTION                              0.053930   \n",
       "HALLMARK_UNFOLDED_PROTEIN_RESPONSE                        0.054696   \n",
       "\n",
       "                                    Rank (simulated)  Percentile (simulated)  \\\n",
       "pathway                                                                        \n",
       "HALLMARK_G2M_CHECKPOINT                         49.0              100.000000   \n",
       "HALLMARK_MYC_TARGETS_V1                         49.0              100.000000   \n",
       "HALLMARK_E2F_TARGETS                            49.0              100.000000   \n",
       "HALLMARK_MYC_TARGETS_V2                         47.0               95.833333   \n",
       "HALLMARK_MTORC1_SIGNALING                       46.0               93.750000   \n",
       "HALLMARK_DNA_REPAIR                             45.0               91.666667   \n",
       "HALLMARK_OXIDATIVE_PHOSPHORYLATION              44.0               89.583333   \n",
       "HALLMARK_INFLAMMATORY_RESPONSE                  43.0               87.500000   \n",
       "HALLMARK_ALLOGRAFT_REJECTION                    42.0               85.416667   \n",
       "HALLMARK_UNFOLDED_PROTEIN_RESPONSE              41.0               83.333333   \n",
       "\n",
       "                                    Mean padj (simulated)  \\\n",
       "pathway                                                     \n",
       "HALLMARK_G2M_CHECKPOINT                          0.010225   \n",
       "HALLMARK_MYC_TARGETS_V1                          0.011698   \n",
       "HALLMARK_E2F_TARGETS                             0.007393   \n",
       "HALLMARK_MYC_TARGETS_V2                          0.071022   \n",
       "HALLMARK_MTORC1_SIGNALING                        0.135693   \n",
       "HALLMARK_DNA_REPAIR                              0.159171   \n",
       "HALLMARK_OXIDATIVE_PHOSPHORYLATION               0.248496   \n",
       "HALLMARK_INFLAMMATORY_RESPONSE                   0.242602   \n",
       "HALLMARK_ALLOGRAFT_REJECTION                     0.224399   \n",
       "HALLMARK_UNFOLDED_PROTEIN_RESPONSE               0.166565   \n",
       "\n",
       "                                    Std deviation (simulated)  \\\n",
       "pathway                                                         \n",
       "HALLMARK_G2M_CHECKPOINT                              0.017949   \n",
       "HALLMARK_MYC_TARGETS_V1                              0.024162   \n",
       "HALLMARK_E2F_TARGETS                                 0.013072   \n",
       "HALLMARK_MYC_TARGETS_V2                              0.205818   \n",
       "HALLMARK_MTORC1_SIGNALING                            0.269983   \n",
       "HALLMARK_DNA_REPAIR                                  0.309746   \n",
       "HALLMARK_OXIDATIVE_PHOSPHORYLATION                   0.407086   \n",
       "HALLMARK_INFLAMMATORY_RESPONSE                       0.369390   \n",
       "HALLMARK_ALLOGRAFT_REJECTION                         0.349003   \n",
       "HALLMARK_UNFOLDED_PROTEIN_RESPONSE                   0.266656   \n",
       "\n",
       "                                    Number of experiments (simulated)  \\\n",
       "pathway                                                                 \n",
       "HALLMARK_G2M_CHECKPOINT                                            25   \n",
       "HALLMARK_MYC_TARGETS_V1                                            25   \n",
       "HALLMARK_E2F_TARGETS                                               25   \n",
       "HALLMARK_MYC_TARGETS_V2                                            25   \n",
       "HALLMARK_MTORC1_SIGNALING                                          25   \n",
       "HALLMARK_DNA_REPAIR                                                25   \n",
       "HALLMARK_OXIDATIVE_PHOSPHORYLATION                                 25   \n",
       "HALLMARK_INFLAMMATORY_RESPONSE                                     22   \n",
       "HALLMARK_ALLOGRAFT_REJECTION                                       25   \n",
       "HALLMARK_UNFOLDED_PROTEIN_RESPONSE                                 25   \n",
       "\n",
       "                                     Z score  \n",
       "pathway                                       \n",
       "HALLMARK_G2M_CHECKPOINT             2.654366  \n",
       "HALLMARK_MYC_TARGETS_V1            -0.363438  \n",
       "HALLMARK_E2F_TARGETS               -0.015574  \n",
       "HALLMARK_MYC_TARGETS_V2            -0.299367  \n",
       "HALLMARK_MTORC1_SIGNALING          -0.219549  \n",
       "HALLMARK_DNA_REPAIR                 0.469907  \n",
       "HALLMARK_OXIDATIVE_PHOSPHORYLATION  0.135960  \n",
       "HALLMARK_INFLAMMATORY_RESPONSE     -0.566707  \n",
       "HALLMARK_ALLOGRAFT_REJECTION        0.793303  \n",
       "HALLMARK_UNFOLDED_PROTEIN_RESPONSE -0.552943  "
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Create intermediate file: \"<local_dir>/gene_summary_table_<col_to_rank_pathways>.tsv\"\n",
    "summary_pathway_ranks = ranking.generate_summary_table(\n",
    "    template_GSEA_stats_filename,\n",
    "    template_GSEA_stats,\n",
    "    simulated_GSEA_summary_stats,\n",
    "    col_to_rank_pathways,\n",
    "    local_dir,\n",
    "    \"pathway\",\n",
    "    params,\n",
    ")\n",
    "\n",
    "summary_pathway_ranks.sort_values(by=\"Rank (simulated)\", ascending=False).head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create `pathway_summary_filename`\n",
    "summary_pathway_ranks.to_csv(pathway_summary_filename, sep=\"\\t\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Compare pathway ranking"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Studies have found that there are some pathways (gene sets) that are more likely to be significantly enriched in DEGs across a wide range of experimental designs. These generic pathways are not necessarily specific to the biological process being studied but instead represents a more systematic change.\n",
    "\n",
    "We want to compare the ability to detect these generic pathways using our method vs those found by [Powers et. al.](https://www.biorxiv.org/content/10.1101/259440v1.full.pdf) publication.  We will use the `Hallmarks_qvalues_GSEAPreranked.csv` file from https://www.synapse.org/#!Synapse:syn11806255 as a reference. The file contains the q-value (adjusted p-value) for the test: given the enrichment score (ES) of the experiment is significant compared to the null distribution of enrichment scores, where the null set is generated from permuted gene sets. For each gene set (pathway) they calculate the q-value using this test.\n",
    "\n",
    "\n",
    "To get a `reference ranking`, we calculate the fraction of experiments that a given pathway was significant (q-value <0.05) and use this rank pathways. `Our ranking` is to rank pathways based on the median q-value across the simulated experiments. We can then compare `our ranking` versus the `reference ranking.`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load Powers et. al. results file\n",
    "powers_rank_filename = os.path.join(\n",
    "    base_dir, dataset_name, \"data\", \"metadata\", \"Hallmarks_qvalues_GSEAPreranked.csv\"\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(50, 443)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>GSE10281_DEG_Expt1_Control_vs_Group1_qval</th>\n",
       "      <th>GSE10299_DEG_Expt1_Control_vs_Group1_qval</th>\n",
       "      <th>GSE10299_DEG_Expt1_Control_vs_Group2_qval</th>\n",
       "      <th>GSE10311_DEG_Expt1_Control_vs_Group1_qval</th>\n",
       "      <th>GSE10410_DEG_Expt1_Control_vs_Group1_qval</th>\n",
       "      <th>GSE10410_DEG_Expt1_Control_vs_Group2_qval</th>\n",
       "      <th>GSE10592_DEG_Expt1_Control_vs_Group1_qval</th>\n",
       "      <th>GSE10592_DEG_Expt2_Control_vs_Group1_qval</th>\n",
       "      <th>GSE11142_DEG_Expt1_Control_vs_Group1_qval</th>\n",
       "      <th>GSE11285_DEG_Expt1_Control_vs_Group1_qval</th>\n",
       "      <th>...</th>\n",
       "      <th>GSE79761_DEG_Expt1_Control_vs_Group2_qval</th>\n",
       "      <th>GSE79761_DEG_Expt1_Control_vs_Group3_qval</th>\n",
       "      <th>GSE83120_DEG_Expt1_Control_vs_Group1_qval</th>\n",
       "      <th>GSE83491_DEG_Expt1_Control_vs_Group1_qval</th>\n",
       "      <th>GSE83491_DEG_Expt2_Control_vs_Group1_qval</th>\n",
       "      <th>GSE8588_DEG_Expt1_Control_vs_Group1_qval</th>\n",
       "      <th>GSE8588_DEG_Expt1_Control_vs_Group2_qval</th>\n",
       "      <th>GSE8597_DEG_Expt1_Control_vs_Group1_qval</th>\n",
       "      <th>GSE8658_DEG_Expt1_Control_vs_Group1_qval</th>\n",
       "      <th>GSE9988_DEG_Expt1_Control_vs_Group1_qval</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Term</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>HALLMARK_APICAL_JUNCTION</th>\n",
       "      <td>0.010497</td>\n",
       "      <td>0.994108</td>\n",
       "      <td>0.045536</td>\n",
       "      <td>0.498151</td>\n",
       "      <td>0.538082</td>\n",
       "      <td>0.429205</td>\n",
       "      <td>0.355439</td>\n",
       "      <td>0.236627</td>\n",
       "      <td>0.767614</td>\n",
       "      <td>0.858065</td>\n",
       "      <td>...</td>\n",
       "      <td>0.637333</td>\n",
       "      <td>0.019621</td>\n",
       "      <td>0.086923</td>\n",
       "      <td>0.453315</td>\n",
       "      <td>0.730539</td>\n",
       "      <td>0.005366</td>\n",
       "      <td>0.863592</td>\n",
       "      <td>0.102867</td>\n",
       "      <td>0.639821</td>\n",
       "      <td>0.465678</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>HALLMARK_APICAL_SURFACE</th>\n",
       "      <td>0.443835</td>\n",
       "      <td>0.598008</td>\n",
       "      <td>0.181818</td>\n",
       "      <td>0.407927</td>\n",
       "      <td>0.130422</td>\n",
       "      <td>0.588732</td>\n",
       "      <td>0.192183</td>\n",
       "      <td>0.003519</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.809823</td>\n",
       "      <td>...</td>\n",
       "      <td>0.260565</td>\n",
       "      <td>0.018686</td>\n",
       "      <td>0.691567</td>\n",
       "      <td>0.588169</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.172392</td>\n",
       "      <td>0.657082</td>\n",
       "      <td>0.008060</td>\n",
       "      <td>0.618102</td>\n",
       "      <td>0.229456</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>HALLMARK_PEROXISOME</th>\n",
       "      <td>0.110897</td>\n",
       "      <td>0.988705</td>\n",
       "      <td>0.118807</td>\n",
       "      <td>0.094770</td>\n",
       "      <td>0.004288</td>\n",
       "      <td>0.115514</td>\n",
       "      <td>0.172297</td>\n",
       "      <td>0.035202</td>\n",
       "      <td>0.682219</td>\n",
       "      <td>0.393990</td>\n",
       "      <td>...</td>\n",
       "      <td>0.447485</td>\n",
       "      <td>0.380294</td>\n",
       "      <td>0.058549</td>\n",
       "      <td>0.110354</td>\n",
       "      <td>0.215365</td>\n",
       "      <td>0.006037</td>\n",
       "      <td>0.034110</td>\n",
       "      <td>0.331770</td>\n",
       "      <td>0.005224</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>HALLMARK_ADIPOGENESIS</th>\n",
       "      <td>0.011177</td>\n",
       "      <td>0.461255</td>\n",
       "      <td>0.171288</td>\n",
       "      <td>0.020412</td>\n",
       "      <td>0.010720</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.191663</td>\n",
       "      <td>0.846586</td>\n",
       "      <td>0.895637</td>\n",
       "      <td>0.056589</td>\n",
       "      <td>...</td>\n",
       "      <td>0.604963</td>\n",
       "      <td>0.805532</td>\n",
       "      <td>0.054415</td>\n",
       "      <td>0.022888</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.735339</td>\n",
       "      <td>0.636778</td>\n",
       "      <td>0.649874</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.435622</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>HALLMARK_ANGIOGENESIS</th>\n",
       "      <td>0.006014</td>\n",
       "      <td>0.380219</td>\n",
       "      <td>0.663985</td>\n",
       "      <td>0.992954</td>\n",
       "      <td>0.175682</td>\n",
       "      <td>0.010256</td>\n",
       "      <td>0.004750</td>\n",
       "      <td>0.015640</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.402218</td>\n",
       "      <td>...</td>\n",
       "      <td>0.007489</td>\n",
       "      <td>0.023545</td>\n",
       "      <td>0.965871</td>\n",
       "      <td>0.035213</td>\n",
       "      <td>0.646400</td>\n",
       "      <td>0.004025</td>\n",
       "      <td>0.209530</td>\n",
       "      <td>0.007165</td>\n",
       "      <td>0.181685</td>\n",
       "      <td>0.108067</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 443 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                          GSE10281_DEG_Expt1_Control_vs_Group1_qval  \\\n",
       "Term                                                                  \n",
       "HALLMARK_APICAL_JUNCTION                                   0.010497   \n",
       "HALLMARK_APICAL_SURFACE                                    0.443835   \n",
       "HALLMARK_PEROXISOME                                        0.110897   \n",
       "HALLMARK_ADIPOGENESIS                                      0.011177   \n",
       "HALLMARK_ANGIOGENESIS                                      0.006014   \n",
       "\n",
       "                          GSE10299_DEG_Expt1_Control_vs_Group1_qval  \\\n",
       "Term                                                                  \n",
       "HALLMARK_APICAL_JUNCTION                                   0.994108   \n",
       "HALLMARK_APICAL_SURFACE                                    0.598008   \n",
       "HALLMARK_PEROXISOME                                        0.988705   \n",
       "HALLMARK_ADIPOGENESIS                                      0.461255   \n",
       "HALLMARK_ANGIOGENESIS                                      0.380219   \n",
       "\n",
       "                          GSE10299_DEG_Expt1_Control_vs_Group2_qval  \\\n",
       "Term                                                                  \n",
       "HALLMARK_APICAL_JUNCTION                                   0.045536   \n",
       "HALLMARK_APICAL_SURFACE                                    0.181818   \n",
       "HALLMARK_PEROXISOME                                        0.118807   \n",
       "HALLMARK_ADIPOGENESIS                                      0.171288   \n",
       "HALLMARK_ANGIOGENESIS                                      0.663985   \n",
       "\n",
       "                          GSE10311_DEG_Expt1_Control_vs_Group1_qval  \\\n",
       "Term                                                                  \n",
       "HALLMARK_APICAL_JUNCTION                                   0.498151   \n",
       "HALLMARK_APICAL_SURFACE                                    0.407927   \n",
       "HALLMARK_PEROXISOME                                        0.094770   \n",
       "HALLMARK_ADIPOGENESIS                                      0.020412   \n",
       "HALLMARK_ANGIOGENESIS                                      0.992954   \n",
       "\n",
       "                          GSE10410_DEG_Expt1_Control_vs_Group1_qval  \\\n",
       "Term                                                                  \n",
       "HALLMARK_APICAL_JUNCTION                                   0.538082   \n",
       "HALLMARK_APICAL_SURFACE                                    0.130422   \n",
       "HALLMARK_PEROXISOME                                        0.004288   \n",
       "HALLMARK_ADIPOGENESIS                                      0.010720   \n",
       "HALLMARK_ANGIOGENESIS                                      0.175682   \n",
       "\n",
       "                          GSE10410_DEG_Expt1_Control_vs_Group2_qval  \\\n",
       "Term                                                                  \n",
       "HALLMARK_APICAL_JUNCTION                                   0.429205   \n",
       "HALLMARK_APICAL_SURFACE                                    0.588732   \n",
       "HALLMARK_PEROXISOME                                        0.115514   \n",
       "HALLMARK_ADIPOGENESIS                                      0.000000   \n",
       "HALLMARK_ANGIOGENESIS                                      0.010256   \n",
       "\n",
       "                          GSE10592_DEG_Expt1_Control_vs_Group1_qval  \\\n",
       "Term                                                                  \n",
       "HALLMARK_APICAL_JUNCTION                                   0.355439   \n",
       "HALLMARK_APICAL_SURFACE                                    0.192183   \n",
       "HALLMARK_PEROXISOME                                        0.172297   \n",
       "HALLMARK_ADIPOGENESIS                                      0.191663   \n",
       "HALLMARK_ANGIOGENESIS                                      0.004750   \n",
       "\n",
       "                          GSE10592_DEG_Expt2_Control_vs_Group1_qval  \\\n",
       "Term                                                                  \n",
       "HALLMARK_APICAL_JUNCTION                                   0.236627   \n",
       "HALLMARK_APICAL_SURFACE                                    0.003519   \n",
       "HALLMARK_PEROXISOME                                        0.035202   \n",
       "HALLMARK_ADIPOGENESIS                                      0.846586   \n",
       "HALLMARK_ANGIOGENESIS                                      0.015640   \n",
       "\n",
       "                          GSE11142_DEG_Expt1_Control_vs_Group1_qval  \\\n",
       "Term                                                                  \n",
       "HALLMARK_APICAL_JUNCTION                                   0.767614   \n",
       "HALLMARK_APICAL_SURFACE                                    1.000000   \n",
       "HALLMARK_PEROXISOME                                        0.682219   \n",
       "HALLMARK_ADIPOGENESIS                                      0.895637   \n",
       "HALLMARK_ANGIOGENESIS                                      1.000000   \n",
       "\n",
       "                          GSE11285_DEG_Expt1_Control_vs_Group1_qval  \\\n",
       "Term                                                                  \n",
       "HALLMARK_APICAL_JUNCTION                                   0.858065   \n",
       "HALLMARK_APICAL_SURFACE                                    0.809823   \n",
       "HALLMARK_PEROXISOME                                        0.393990   \n",
       "HALLMARK_ADIPOGENESIS                                      0.056589   \n",
       "HALLMARK_ANGIOGENESIS                                      0.402218   \n",
       "\n",
       "                                            ...                     \\\n",
       "Term                                        ...                      \n",
       "HALLMARK_APICAL_JUNCTION                    ...                      \n",
       "HALLMARK_APICAL_SURFACE                     ...                      \n",
       "HALLMARK_PEROXISOME                         ...                      \n",
       "HALLMARK_ADIPOGENESIS                       ...                      \n",
       "HALLMARK_ANGIOGENESIS                       ...                      \n",
       "\n",
       "                          GSE79761_DEG_Expt1_Control_vs_Group2_qval  \\\n",
       "Term                                                                  \n",
       "HALLMARK_APICAL_JUNCTION                                   0.637333   \n",
       "HALLMARK_APICAL_SURFACE                                    0.260565   \n",
       "HALLMARK_PEROXISOME                                        0.447485   \n",
       "HALLMARK_ADIPOGENESIS                                      0.604963   \n",
       "HALLMARK_ANGIOGENESIS                                      0.007489   \n",
       "\n",
       "                          GSE79761_DEG_Expt1_Control_vs_Group3_qval  \\\n",
       "Term                                                                  \n",
       "HALLMARK_APICAL_JUNCTION                                   0.019621   \n",
       "HALLMARK_APICAL_SURFACE                                    0.018686   \n",
       "HALLMARK_PEROXISOME                                        0.380294   \n",
       "HALLMARK_ADIPOGENESIS                                      0.805532   \n",
       "HALLMARK_ANGIOGENESIS                                      0.023545   \n",
       "\n",
       "                          GSE83120_DEG_Expt1_Control_vs_Group1_qval  \\\n",
       "Term                                                                  \n",
       "HALLMARK_APICAL_JUNCTION                                   0.086923   \n",
       "HALLMARK_APICAL_SURFACE                                    0.691567   \n",
       "HALLMARK_PEROXISOME                                        0.058549   \n",
       "HALLMARK_ADIPOGENESIS                                      0.054415   \n",
       "HALLMARK_ANGIOGENESIS                                      0.965871   \n",
       "\n",
       "                          GSE83491_DEG_Expt1_Control_vs_Group1_qval  \\\n",
       "Term                                                                  \n",
       "HALLMARK_APICAL_JUNCTION                                   0.453315   \n",
       "HALLMARK_APICAL_SURFACE                                    0.588169   \n",
       "HALLMARK_PEROXISOME                                        0.110354   \n",
       "HALLMARK_ADIPOGENESIS                                      0.022888   \n",
       "HALLMARK_ANGIOGENESIS                                      0.035213   \n",
       "\n",
       "                          GSE83491_DEG_Expt2_Control_vs_Group1_qval  \\\n",
       "Term                                                                  \n",
       "HALLMARK_APICAL_JUNCTION                                   0.730539   \n",
       "HALLMARK_APICAL_SURFACE                                    1.000000   \n",
       "HALLMARK_PEROXISOME                                        0.215365   \n",
       "HALLMARK_ADIPOGENESIS                                      0.000000   \n",
       "HALLMARK_ANGIOGENESIS                                      0.646400   \n",
       "\n",
       "                          GSE8588_DEG_Expt1_Control_vs_Group1_qval  \\\n",
       "Term                                                                 \n",
       "HALLMARK_APICAL_JUNCTION                                  0.005366   \n",
       "HALLMARK_APICAL_SURFACE                                   0.172392   \n",
       "HALLMARK_PEROXISOME                                       0.006037   \n",
       "HALLMARK_ADIPOGENESIS                                     0.735339   \n",
       "HALLMARK_ANGIOGENESIS                                     0.004025   \n",
       "\n",
       "                          GSE8588_DEG_Expt1_Control_vs_Group2_qval  \\\n",
       "Term                                                                 \n",
       "HALLMARK_APICAL_JUNCTION                                  0.863592   \n",
       "HALLMARK_APICAL_SURFACE                                   0.657082   \n",
       "HALLMARK_PEROXISOME                                       0.034110   \n",
       "HALLMARK_ADIPOGENESIS                                     0.636778   \n",
       "HALLMARK_ANGIOGENESIS                                     0.209530   \n",
       "\n",
       "                          GSE8597_DEG_Expt1_Control_vs_Group1_qval  \\\n",
       "Term                                                                 \n",
       "HALLMARK_APICAL_JUNCTION                                  0.102867   \n",
       "HALLMARK_APICAL_SURFACE                                   0.008060   \n",
       "HALLMARK_PEROXISOME                                       0.331770   \n",
       "HALLMARK_ADIPOGENESIS                                     0.649874   \n",
       "HALLMARK_ANGIOGENESIS                                     0.007165   \n",
       "\n",
       "                          GSE8658_DEG_Expt1_Control_vs_Group1_qval  \\\n",
       "Term                                                                 \n",
       "HALLMARK_APICAL_JUNCTION                                  0.639821   \n",
       "HALLMARK_APICAL_SURFACE                                   0.618102   \n",
       "HALLMARK_PEROXISOME                                       0.005224   \n",
       "HALLMARK_ADIPOGENESIS                                     0.000000   \n",
       "HALLMARK_ANGIOGENESIS                                     0.181685   \n",
       "\n",
       "                          GSE9988_DEG_Expt1_Control_vs_Group1_qval  \n",
       "Term                                                                \n",
       "HALLMARK_APICAL_JUNCTION                                  0.465678  \n",
       "HALLMARK_APICAL_SURFACE                                   0.229456  \n",
       "HALLMARK_PEROXISOME                                       1.000000  \n",
       "HALLMARK_ADIPOGENESIS                                     0.435622  \n",
       "HALLMARK_ANGIOGENESIS                                     0.108067  \n",
       "\n",
       "[5 rows x 443 columns]"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Read Powers et. al. data\n",
    "# This file contains qvalue results for hallmark pathways across ~400 experiments\n",
    "powers_rank_df = pd.read_csv(powers_rank_filename, header=0, index_col=0)\n",
    "powers_rank_df.drop([\"Category\"], axis=1, inplace=True)\n",
    "print(powers_rank_df.shape)\n",
    "powers_rank_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Fraction enriched</th>\n",
       "      <th>Powers Rank</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Term</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>HALLMARK_TNFA_SIGNALING_VIA_NFKB</th>\n",
       "      <td>0.668172</td>\n",
       "      <td>50.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>HALLMARK_E2F_TARGETS</th>\n",
       "      <td>0.645598</td>\n",
       "      <td>49.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>HALLMARK_G2M_CHECKPOINT</th>\n",
       "      <td>0.586907</td>\n",
       "      <td>48.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>HALLMARK_MITOTIC_SPINDLE</th>\n",
       "      <td>0.539503</td>\n",
       "      <td>47.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>HALLMARK_MYC_TARGETS_V1</th>\n",
       "      <td>0.528217</td>\n",
       "      <td>46.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                  Fraction enriched  Powers Rank\n",
       "Term                                                            \n",
       "HALLMARK_TNFA_SIGNALING_VIA_NFKB           0.668172         50.0\n",
       "HALLMARK_E2F_TARGETS                       0.645598         49.0\n",
       "HALLMARK_G2M_CHECKPOINT                    0.586907         48.0\n",
       "HALLMARK_MITOTIC_SPINDLE                   0.539503         47.0\n",
       "HALLMARK_MYC_TARGETS_V1                    0.528217         46.0"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Count the number of experiments where a given pathway was found to be enriched (qvalue < 0.05)\n",
    "total_num_experiments = powers_rank_df.shape[1]\n",
    "frac_enriched_pathways = (powers_rank_df < 0.05).sum(axis=1) / total_num_experiments\n",
    "\n",
    "# Rank pathways from 0-50, 50 indicating that the pathways was frequently enriched\n",
    "pathway_ranks = frac_enriched_pathways.rank()\n",
    "\n",
    "powers_rank_stats_df = pd.DataFrame(\n",
    "    data={\n",
    "        \"Fraction enriched\": frac_enriched_pathways.values,\n",
    "        \"Powers Rank\": pathway_ranks.values,\n",
    "    },\n",
    "    index=powers_rank_df.index,\n",
    ")\n",
    "powers_rank_stats_df.sort_values(by=\"Powers Rank\", ascending=False).head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save reference file for input into comparison\n",
    "powers_rank_processed_filename = os.path.join(\n",
    "    base_dir,\n",
    "    dataset_name,\n",
    "    \"data\",\n",
    "    \"metadata\",\n",
    "    \"Hallmarks_qvalues_GSEAPreranked_processed.tsv\",\n",
    ")\n",
    "\n",
    "powers_rank_stats_df.to_csv(\n",
    "    powers_rank_processed_filename,\n",
    "    sep=\"\\t\",\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "r = 0.6190464797851615\n",
      "p = 1.6498645355243548e-06\n",
      "ci_low = 0.4011545199533903\n",
      "ci_high = 0.7933458529105532\n",
      "{'r': 0.6190464797851615, 'p': 1.6498645355243548e-06, 'ci_low': 0.4011545199533903, 'ci_high': 0.7933458529105532}\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'r': 0.6190464797851615,\n",
       " 'p': 1.6498645355243548e-06,\n",
       " 'ci_low': 0.4011545199533903,\n",
       " 'ci_high': 0.7933458529105532}"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "figure_filename = f\"pathway_ranking_{col_to_rank_pathways}.svg\"\n",
    "\n",
    "ranking.compare_pathway_ranking(\n",
    "    summary_pathway_ranks, powers_rank_processed_filename, figure_filename\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Takeaway:**\n",
    "* Our method ranked pathways using median adjusted p-value score across simulated experiments.\n",
    "* Powers et. al. ranked pathways based on the fraction of experiments they had adjusted p-value < 0.05.\n",
    "* Here we validated that our analysis pipeline is working correctly by comparing pathway ranks obtained from a (Powers et. al.)-trained VAE model vs pathway ranking based on manual curation using Powers et. al datasets. We expect to see a high correlation between pathway ranks given that we are using the same training dataset. Indeed that is what we find."
   ]
  }
 ],
 "metadata": {
  "jupytext": {
   "encoding": "# -*- coding: utf-8 -*-",
   "formats": "ipynb,py"
  },
  "kernelspec": {
   "display_name": "Python [conda env:generic_expression_new]",
   "language": "python",
   "name": "conda-env-generic_expression_new-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}