analysis/genie/variant_analysis/genie_search_analysis.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "34b1fa10-28ad-44e4-b8a4-49aeef14b03e",
   "metadata": {},
   "source": [
    "# GENIE Search Analysis\n",
    "\n",
    "This notebook performs set operations to determine matches between normalized variants from GENIE and normalized variants from CIViC, Molecular Oncology Almanac, and ClinVar."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8c7e6ac4-3e80-4abb-9b9c-8de41518e405",
   "metadata": {},
   "source": [
    "### Prerequisites\n",
    "The following notebook must be run before running this analysis: \n",
    "\n",
    "- `genie/pre_variant_analysis/genie_pre_variant_analysis.ipynb`\n",
    "\n",
    "This notebook uses `data_mutations_extended.txt` from Synapse. You will need to create an account to download data from [here](https://www.synapse.org/#!Synapse:syn51355986). This notebook expects the `data_mutations_extended.txt` to be in the `pre_variant_analysis` directory."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "8d53de2a-4620-4700-b34b-56035b651ffb",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import relevant libraries\n",
    "import ndjson\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from pathlib import Path\n",
    "import gzip"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8aaa6a3f-8b65-4383-8f7a-4c72baff9487",
   "metadata": {},
   "source": [
    "## Load Data (CIViC, Molecular Oncology Almanac, ClinVar, GENIE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "22709d08-246b-4d37-a66f-f686f719cb1e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load normalized variants for CIViC\n",
    "civic_variants = pd.read_csv(\n",
    "    \"../../civic/variation_analysis/able_to_normalize_queries.csv\", sep=\"\\t\"\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "f8f40171-dcba-41c1-b464-26af010fd526",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load normalized variants for MOA\n",
    "moa_variants = pd.read_csv(\n",
    "    \"../../moa/feature_analysis/able_to_normalize_queries.csv\", sep=\"\\t\"\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "94c89830",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load normalized variants for ClinVar\n",
    "with gzip.open(\"../../clinvar/output-variation_identity-vrs-1.3.ndjson.gz\", \"rb\") as f:\n",
    "    records = ndjson.load(f)\n",
    "\n",
    "clinvar_variants = pd.json_normalize(records)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "09ed1f43",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1432052, 12)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Get GENIE variant data\n",
    "genie_variants_df = pd.read_csv(\n",
    "    \"../pre_variant_analysis/variation_normalizer_output/genie_variants_df.csv\",\n",
    "    sep=\"\\t\",\n",
    ")\n",
    "genie_variants_df.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ed00e26a-936e-4761-9018-ba2971857215",
   "metadata": {},
   "source": [
    "## Normalized Genomic and Protein GENIE variants \n",
    "Due to GENIE licenses, please run the analysis notebook in `genie/pre_variant_analysis` to generate the datasets below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "c1efda24-dd64-42de-9e27-1d76d9fd1bdb",
   "metadata": {},
   "outputs": [],
   "source": [
    "genie_genomic_normalized = pd.read_csv(\n",
    "    \"../pre_variant_analysis/variation_normalizer_output/able_to_normalize_genomic_queries.csv\",\n",
    "    sep=\"\\t\",\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "45efc26f-f865-4a4d-92c8-1e8d29dcf5e5",
   "metadata": {},
   "outputs": [],
   "source": [
    "genie_protein_normalized = pd.read_csv(\n",
    "    \"../pre_variant_analysis/variation_normalizer_output/able_to_normalize_protein_queries.csv\",\n",
    "    sep=\"\\t\",\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "16e14634-84ab-428c-a86b-784c49088094",
   "metadata": {},
   "source": [
    "## Create Output Directory"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "9b0d9d3e-0b79-4b5e-8db8-5a13de17dbdb",
   "metadata": {},
   "outputs": [],
   "source": [
    "path = Path(\"variant_analysis_output\")\n",
    "path.mkdir(exist_ok=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2324a30e-250a-45b6-a3bf-a023e855a92f",
   "metadata": {},
   "source": [
    "## Variant-Level Analysis\n",
    "In this section we perform set intersections to determine how many normalized variants from GENIE match to MOA, CIViC, and ClinVar."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "1db857a4-834a-458b-a9f9-d5fb19db32ea",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Helper functions\n",
    "def create_var_patient_data(filename: str, data: list) -> None:\n",
    "    \"\"\"Write variant/patient data to a file\n",
    "    :param filename: The file to write to\n",
    "    :param data: A list of variants or patients\n",
    "    \"\"\"\n",
    "    with open(filename, \"w\") as f:\n",
    "        for match in data:\n",
    "            f.write(match + \"\\n\")\n",
    "\n",
    "\n",
    "def set_intersection(genie: list, kb: list) -> set:\n",
    "    \"\"\"Intersect GENIE variants with variants from different knowledgebases\n",
    "    :param genie: A list of GENIE variants\n",
    "    :param kb: A list of variants from a knowledgebase\n",
    "    :return: set of common variants\n",
    "    \"\"\"\n",
    "    return set(genie).intersection(kb)\n",
    "\n",
    "\n",
    "def var_analysis(genie: list, kb: list, filename: str) -> tuple:\n",
    "    \"\"\"Perform set intersection, write to file for variant analysis\n",
    "    :param genie: A list of GENIE variants\n",
    "    :param kb: A list of variants from a knowledgebase\n",
    "    :param filename: The file to write to\n",
    "    :return: A tuple containing a list of shared variants, and the length of the list\n",
    "    \"\"\"\n",
    "    match = set_intersection(genie, kb)\n",
    "    create_var_patient_data(filename, list(match))\n",
    "    return match, len(match)\n",
    "\n",
    "\n",
    "def is_var_in_kb(var: str, kb: set) -> bool:\n",
    "    \"\"\"Check if a variant is in a knowledgebase\n",
    "    :param var: A variant\n",
    "    :param kb: A set of variants\n",
    "    :return: True if variant is in knowledgebase\n",
    "    \"\"\"\n",
    "    return var in kb"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b85922d3-a86a-4f6b-9b32-cffde986a0bb",
   "metadata": {},
   "source": [
    "### MOA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "85250b46",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'The number of matched protein variants in MOA is: 122'"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Intersect normalized GENIE protein variants with normalized MOA protein variants\n",
    "genie_moa_match = var_analysis(\n",
    "    genie_protein_normalized[\"vrs_id\"].to_list(),\n",
    "    set(moa_variants[\"vrs_id\"].to_list()),\n",
    "    \"variant_analysis_output/genie_moa_match.txt\",\n",
    ")\n",
    "f\"The number of matched protein variants in MOA is: {genie_moa_match[1]}\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0e6d3a50-752d-4ffd-aa17-25e785ff9da8",
   "metadata": {},
   "source": [
    "### CIViC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "a531e9cc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'The number of matched genomic variants in CIViC is: 203'"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Intersect normalized GENIE genomic variants with normalized CIViC variants\n",
    "genie_civic_match_genomic = var_analysis(\n",
    "    genie_genomic_normalized[\"vrs_id\"].to_list(),\n",
    "    set(civic_variants[\"vrs_id\"].to_list()),\n",
    "    \"variant_analysis_output/genie_civic_match_genomic.txt\",\n",
    ")\n",
    "f\"The number of matched genomic variants in CIViC is: {genie_civic_match_genomic[1]}\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "253da819",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'The number of matched protein variants in CIViC is: 967'"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Intersect normalized GENIE protein variants with normalized CIViC variants\n",
    "genie_civic_match_protein = var_analysis(\n",
    "    genie_protein_normalized[\"vrs_id\"].to_list(),\n",
    "    set(civic_variants[\"vrs_id\"].to_list()),\n",
    "    \"variant_analysis_output/genie_civic_match_protein.txt\",\n",
    ")\n",
    "f\"The number of matched protein variants in CIViC is: {genie_civic_match_protein[1]}\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "e652af20-66da-4b52-bcc7-1587b518a7a8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'The number of matched variants in CIViC is: 1170'"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Inspect union of GENIE genomic and protein variants\n",
    "civic_genie_union = genie_civic_match_genomic[0].union(genie_civic_match_protein[0])\n",
    "create_var_patient_data(\n",
    "    \"variant_analysis_output/genie_civic_match_union.txt\", civic_genie_union\n",
    ")\n",
    "f\"The number of matched variants in CIViC is: {len(civic_genie_union)}\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "faa0c4c4-9a55-4623-b86b-d71a72cf5d3e",
   "metadata": {},
   "source": [
    "### ClinVar"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "be1c454e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'The number of matched genomic variants in ClinVar is: 94300'"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Intersect normalized GENIE genomic variants with normalized ClinVar variants\n",
    "genie_clinvar_match_genomic = var_analysis(\n",
    "    genie_genomic_normalized[\"vrs_id\"].to_list(),\n",
    "    set(clinvar_variants[\"out._id\"].to_list()),\n",
    "    \"variant_analysis_output/genie_clinvar_match_genomic.txt\",\n",
    ")\n",
    "f\"The number of matched genomic variants in ClinVar is: {genie_clinvar_match_genomic[1]}\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "e88ff7f4-fca8-43d8-a683-014c768324c6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'The number of matched protein variants in ClinVar is: 0'"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Intersect normalized GENIE protein variants with normalized ClinVar variants\n",
    "genie_clinvar_match_protein = set_intersection(\n",
    "    genie_protein_normalized[\"vrs_id\"].to_list(),\n",
    "    set(clinvar_variants[\"out._id\"].to_list()),\n",
    ")\n",
    "f\"The number of matched protein variants in ClinVar is: {len(genie_clinvar_match_protein)}\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "69cc2b8e-7ab0-43be-89af-b908339bbb11",
   "metadata": {},
   "source": [
    "## Patient-Level Analysis\n",
    "In this section we summarize variant matching at the patient level, with >= 1 shared variant counting as a match."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f457f72b-1875-4ffe-9c84-f9f237fd3d2b",
   "metadata": {},
   "source": [
    "### Filter to include normalized genomic and protein variants"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "85cb5569",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Helper function - create dataframes with vrs_ids for genomic and protein variants\n",
    "def get_vrs_ids(df: pd.DataFrame, col_name: str) -> pd.DataFrame:\n",
    "    \"\"\"Given a list of GENIE variants, add a column containing the VRS IDs\n",
    "    :param df: A dataframe containing the VRS IDs for each variant\n",
    "    :param col_name: Indicates if genomic or protein variants will be added\n",
    "    :return: The updated genie_variants dataframe\n",
    "    \"\"\"\n",
    "    genie_variants_mapping = {}\n",
    "    for row in df.itertuples(index=False):\n",
    "        if row.query not in genie_variants_mapping:\n",
    "            genie_variants_mapping[row.query] = row.vrs_id\n",
    "\n",
    "    queries = genie_variants_df[col_name].to_list()\n",
    "    vrs_ids_list = [genie_variants_mapping.get(query, np.nan) for query in queries]\n",
    "\n",
    "    key = \"vrs_id_genomic\" if col_name == \"coordinates\" else \"vrs_id_protein\"\n",
    "    genie_variants_df[key] = vrs_ids_list\n",
    "    return genie_variants_df[genie_variants_df[key].notna()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "923bd9aa",
   "metadata": {},
   "outputs": [],
   "source": [
    "genie_variants_genomic = get_vrs_ids(genie_genomic_normalized, \"coordinates\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "b6a8a935",
   "metadata": {},
   "outputs": [],
   "source": [
    "genie_variants_protein = get_vrs_ids(genie_protein_normalized, \"free_text_p_short\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "083cce45-4253-46b0-8f5b-13d2bd5f24fa",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Add columns indicating which variants are in which knowledgebase\n",
    "genie_variants_df[\"in_civic_genomic\"] = genie_variants_df[\"vrs_id_genomic\"].map(\n",
    "    lambda x: is_var_in_kb(x, genie_civic_match_genomic[0])\n",
    ")\n",
    "genie_variants_df[\"in_clinvar\"] = genie_variants_df[\"vrs_id_genomic\"].map(\n",
    "    lambda x: is_var_in_kb(x, genie_clinvar_match_genomic[0])\n",
    ")\n",
    "genie_variants_df[\"in_civic_protein\"] = genie_variants_df[\"vrs_id_protein\"].map(\n",
    "    lambda x: is_var_in_kb(x, genie_civic_match_protein[0])\n",
    ")\n",
    "genie_variants_df[\"in_moa\"] = genie_variants_df[\"vrs_id_protein\"].map(\n",
    "    lambda x: is_var_in_kb(x, genie_moa_match[0])\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "3a440a0e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Helper function - create patient dictionary\n",
    "def get_patient_vars(df: pd.DataFrame, var_type: str) -> dict:\n",
    "    \"\"\"Create a dictionary reporting the variants a patient has\n",
    "    :param df: A dataframe with GENIE genomic or protein variants\n",
    "    :param var_type: Genomic or protein\n",
    "    :return: A dictionary of patient variants\n",
    "    \"\"\"\n",
    "    patient_dict = dict()\n",
    "    for row in df.itertuples(index=False):\n",
    "        key = row.Tumor_Sample_Barcode\n",
    "        var = row.vrs_id_genomic if var_type == \"genomic\" else row.vrs_id_protein\n",
    "        if key in patient_dict:\n",
    "            curr = patient_dict[key]\n",
    "            if not pd.isna(var):\n",
    "                curr.append(var)\n",
    "            patient_dict[key] = curr\n",
    "        else:\n",
    "            if not pd.isna(var):\n",
    "                patient_dict[key] = [var]\n",
    "    return patient_dict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "b2424f72",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create genomic variants patient dictionary\n",
    "pvd_genomic = get_patient_vars(genie_variants_genomic, \"genomic\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "d684ac66",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create protein variants patient dictionary\n",
    "pvd_protein = get_patient_vars(genie_variants_protein, \"protein\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "591ff1c0",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Patient Analysis Helper Function\n",
    "def patient_matching(patient_dict: dict, var_set: set, filename: str) -> tuple:\n",
    "    \"\"\"Perform patient matching for a given source\n",
    "    :param patient_dict: A dictionary of patient variants\n",
    "    :param var_set: A set of common variants\n",
    "    :param filename: The file to write to\n",
    "    :return: A tuple containing a list of patients with common variants, and the number of patients in that list\n",
    "    \"\"\"\n",
    "    count = 0\n",
    "    patients = []\n",
    "    for key in patient_dict:\n",
    "        isect = set(patient_dict[key]).intersection(var_set)\n",
    "        if isect:\n",
    "            count += 1\n",
    "            patients.append(key)\n",
    "    create_var_patient_data(filename, patients)\n",
    "    return count, patients"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f4ee515b-7c6d-4084-9f52-9fdab03242f0",
   "metadata": {},
   "source": [
    "### MOA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "2fb93337",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'There are 41828 patients with matching protein variant data in MOA'"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "moa_match = patient_matching(\n",
    "    pvd_protein, genie_moa_match[0], \"variant_analysis_output/moa_patient_match.txt\"\n",
    ")\n",
    "f\"There are {moa_match[0]} patients with matching protein variant data in MOA\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "aeb32052-4924-4f51-96c5-33c3b99446a0",
   "metadata": {},
   "source": [
    "### CIViC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "37306871",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'There are 84861 patients with matching protein variant data in CIViC'"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "civic_protein_match = patient_matching(\n",
    "    pvd_protein,\n",
    "    genie_civic_match_protein[0],\n",
    "    \"variant_analysis_output/civic_protein_patient_match.txt\",\n",
    ")\n",
    "f\"There are {civic_protein_match[0]} patients with matching protein variant data in CIViC\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "a49cdbbc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'There are 853 patients with matching genomic variant data in CIViC'"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "civic_genomic_match = patient_matching(\n",
    "    pvd_genomic,\n",
    "    genie_civic_match_genomic[0],\n",
    "    \"variant_analysis_output/civic_genomic_patient_match.txt\",\n",
    ")\n",
    "f\"There are {civic_genomic_match[0]} patients with matching genomic variant data in CIViC\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "41714c8d-816f-4c35-853f-4bc8807135da",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "272"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Compute number of patients with both genomic and protein variant overlap\n",
    "len(set(civic_genomic_match[1]).intersection(set(civic_protein_match[1])))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c42996b2-d106-4dfe-9239-760be7dbf657",
   "metadata": {},
   "source": [
    "### ClinVar"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "8617b67f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'There are 127103 patients with matching genomic variant data in ClinVar'"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "genie_clinvar_patient_match = patient_matching(\n",
    "    pvd_genomic,\n",
    "    genie_clinvar_match_genomic[0],\n",
    "    \"variant_analysis_output/clinvar_patient_match.txt\",\n",
    ")\n",
    "f\"There are {genie_clinvar_patient_match[0]} patients with matching genomic variant data in ClinVar\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fa61dba5-1170-438e-9b80-d2fb172be3a8",
   "metadata": {},
   "source": [
    "## Variant Counts per Patient\n",
    "Compute the average number of genomic and protein variants per patient."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "99fcb9db",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_avg_vars_per_patient(var_dict: dict) -> float:\n",
    "    \"\"\"Return the average number of variants per patient\n",
    "    :param var_dict: A dictionary of patient variants\n",
    "    :return: The average number of normalized variants\n",
    "    \"\"\"\n",
    "    counts = [len(var_dict[k]) for k in var_dict]\n",
    "    return sum(counts) / len(counts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "4b951a3e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "8.703645296719163"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Normalized average, genomic\n",
    "get_avg_vars_per_patient(pvd_genomic)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "a2f29635",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7.9632307941219915"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Normalized average, protein\n",
    "get_avg_vars_per_patient(pvd_protein)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "879fb2c0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "8.860833124843634"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Average number of normalized variants per patient\n",
    "normalized_variants_df = genie_variants_df[\n",
    "    ~genie_variants_df[\"vrs_id_genomic\"].isna()\n",
    "    | ~genie_variants_df[\"vrs_id_protein\"].isna()\n",
    "]\n",
    "group_by_tsb = list(normalized_variants_df.groupby(\"Tumor_Sample_Barcode\").size())\n",
    "sum(group_by_tsb) / len(group_by_tsb)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "c2c79bbc-c892-4378-9724-77e5af53f1a5",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_kb_avg_vars_per_patient(df: pd.DataFrame) -> float:\n",
    "    \"\"\"Compute the average number of variants per patient given a knowledgebase\n",
    "    :param df: The filtered dataframe of patient variants\n",
    "    :return: The average number of variants\n",
    "    \"\"\"\n",
    "    n_variants = list(df.groupby(\"Tumor_Sample_Barcode\").size())\n",
    "    return round(sum(n_variants) / len(n_variants), 2), len(n_variants)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "6df32553",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "9.75"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Average number of total variants per patient\n",
    "get_kb_avg_vars_per_patient(genie_variants_df)[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "cab7d8e2-c722-43ae-9a3b-206cd0fc2662",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.01"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Normalized matched genomic variants per patient in CIViC\n",
    "genie_df_civic_genomic = genie_variants_df[genie_variants_df[\"in_civic_genomic\"]]\n",
    "civic_genomic_match = get_kb_avg_vars_per_patient(genie_df_civic_genomic)\n",
    "civic_genomic_match[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "47fd7ee3-549b-4c4e-bc83-820ca6fd56d9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.38"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Normalized matched protein variants per patient in CIViC\n",
    "genie_df_civic_protein = genie_variants_df[genie_variants_df[\"in_civic_protein\"]]\n",
    "civic_protein_match = get_kb_avg_vars_per_patient(genie_df_civic_protein)\n",
    "civic_protein_match[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "17b92b3e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.38"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Normalized matched patient variants across CIViC\n",
    "genie_df_civic = genie_variants_df[\n",
    "    genie_variants_df[\"in_civic_genomic\"] | genie_variants_df[\"in_civic_protein\"]\n",
    "]\n",
    "civic_match = get_kb_avg_vars_per_patient(genie_df_civic)\n",
    "civic_match[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "708f2751-b1de-4f46-89db-b2a5a36bb3e7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.1"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Normalized matched protein variants per patient in MOA\n",
    "genie_df_moa = genie_variants_df[genie_variants_df[\"in_moa\"]]\n",
    "moa_protein_match = get_kb_avg_vars_per_patient(genie_df_moa)\n",
    "moa_protein_match[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "504863b3-41d2-4efb-990f-996d876ed9a8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3.22"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Normalized matched genomic variants per patient in ClinVar\n",
    "genie_df_clinvar = genie_variants_df[genie_variants_df[\"in_clinvar\"]]\n",
    "clinvar_genomic_match = get_kb_avg_vars_per_patient(genie_df_clinvar)\n",
    "clinvar_genomic_match[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "504863b3-41d2-4efb-990f-996d876ed9a8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3.22"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Normalized matching across knowlegebases, genomic\n",
    "genie_df_genomic_combined = genie_variants_df[\n",
    "    genie_variants_df[\"in_civic_genomic\"] | genie_variants_df[\"in_clinvar\"]\n",
    "]\n",
    "genomic_combined_match = get_kb_avg_vars_per_patient(genie_df_genomic_combined)\n",
    "genomic_combined_match[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "1913e698-1fa6-4946-ba35-a9e53193d401",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.39"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Normalized matching across knowlegebases, protein\n",
    "genie_df_protein_combined = genie_variants_df[\n",
    "    genie_variants_df[\"in_civic_protein\"] | genie_variants_df[\"in_moa\"]\n",
    "]\n",
    "protein_combined_match = get_kb_avg_vars_per_patient(genie_df_protein_combined)\n",
    "protein_combined_match[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "ba47c5e1-18ac-44a3-b6c3-ea85b20dadfe",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3.25"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Normalized matching across all knowledgebases\n",
    "genie_df_filtered = genie_variants_df[\n",
    "    genie_variants_df[\"in_civic_protein\"]\n",
    "    | genie_variants_df[\"in_moa\"]\n",
    "    | genie_variants_df[\"in_civic_genomic\"]\n",
    "    | genie_variants_df[\"in_clinvar\"]\n",
    "]\n",
    "filtered_match = get_kb_avg_vars_per_patient(genie_df_filtered)\n",
    "filtered_match[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2f1a49a4",
   "metadata": {},
   "source": [
    "## Summary Statistics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "169acd38",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "146857"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Number of unique patients in GENIE dataframe\n",
    "total_patients = genie_variants_df[\"Tumor_Sample_Barcode\"].nunique()\n",
    "total_patients"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "f3af6d8e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "143892"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Number of patients with at least 1 normalized variant\n",
    "normalized_patients = normalized_variants_df[\"Tumor_Sample_Barcode\"].nunique()\n",
    "normalized_patients"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "f4ae7bd7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "128397"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Number of patients with at least 1 normalized variant and a match to a knowledgebase\n",
    "matching_patients = genie_df_filtered[\"Tumor_Sample_Barcode\"].nunique()\n",
    "matching_patients"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "b171f3fc",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Table 1 - Variant Counts\n",
    "summary_df_var = pd.DataFrame(columns=[\"Protein\", \"Genomic\"])\n",
    "summary_df_var.loc[\"# of Unique GENIE Variants\"] = [\n",
    "    genie_variants_df[\"free_text_p_short\"].nunique(),\n",
    "    genie_variants_df[\"coordinates\"].nunique(),\n",
    "]\n",
    "summary_df_var.loc[\"# of Able to Normalize GENIE Queries\"] = [\n",
    "    len(genie_protein_normalized.index),\n",
    "    len(genie_genomic_normalized.index),\n",
    "]\n",
    "summary_df_var.loc[\"# of Unique GENIE VRS IDs\"] = [\n",
    "    normalized_variants_df[\"vrs_id_protein\"].nunique(),\n",
    "    normalized_variants_df[\"vrs_id_genomic\"].nunique(),\n",
    "]\n",
    "summary_df_var.loc[\"# of Unique Normalized GENIE Variants in CIViC\"] = [\n",
    "    genie_df_civic_protein[\"vrs_id_protein\"].nunique(),\n",
    "    genie_df_civic_genomic[\"vrs_id_genomic\"].nunique(),\n",
    "]\n",
    "summary_df_var.loc[\"# of Unique Normalized Protein GENIE Variants in MOA\"] = [\n",
    "    genie_df_moa[\"vrs_id_protein\"].nunique(),\n",
    "    0,\n",
    "]\n",
    "summary_df_var.loc[\"# of Unique Normalized Genomic GENIE Variants in ClinVar\"] = [\n",
    "    0,\n",
    "    genie_df_clinvar[\"vrs_id_genomic\"].nunique(),\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "00e82c3e",
   "metadata": {},
   "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>Protein</th>\n",
       "      <th>Genomic</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th># of Unique GENIE Variants</th>\n",
       "      <td>662149</td>\n",
       "      <td>647541</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th># of Able to Normalize GENIE Queries</th>\n",
       "      <td>566804</td>\n",
       "      <td>646946</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th># of Unique GENIE VRS IDs</th>\n",
       "      <td>565824</td>\n",
       "      <td>646696</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th># of Unique Normalized GENIE Variants in CIViC</th>\n",
       "      <td>967</td>\n",
       "      <td>203</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th># of Unique Normalized Protein GENIE Variants in MOA</th>\n",
       "      <td>122</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th># of Unique Normalized Genomic GENIE Variants in ClinVar</th>\n",
       "      <td>0</td>\n",
       "      <td>94300</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                    Protein  Genomic\n",
       "# of Unique GENIE Variants                           662149   647541\n",
       "# of Able to Normalize GENIE Queries                 566804   646946\n",
       "# of Unique GENIE VRS IDs                            565824   646696\n",
       "# of Unique Normalized GENIE Variants in CIViC          967      203\n",
       "# of Unique Normalized Protein GENIE Variants i...      122        0\n",
       "# of Unique Normalized Genomic GENIE Variants i...        0    94300"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "summary_df_var"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "efc9e0ad",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Helper function\n",
    "def compute_percent(n: int) -> float:\n",
    "    \"\"\"Format percentage\n",
    "    :param n: An integer\n",
    "    :return: A float reporting the percentage\"\"\"\n",
    "    return round(100 * n / total_patients, 1)\n",
    "\n",
    "\n",
    "# Table 2 - Patient Matching. The percentages are computed using the total number of patients in the GENIE dataset.\n",
    "summary_df_patient = pd.DataFrame(columns=[\"Protein\", \"Genomic\", \"Protein or Genomic\"])\n",
    "summary_df_patient.loc[\"All Knowledgebases\"] = [\n",
    "    compute_percent(protein_combined_match[1]),\n",
    "    compute_percent(genomic_combined_match[1]),\n",
    "    compute_percent(filtered_match[1]),\n",
    "]\n",
    "summary_df_patient.loc[\"MOAlmanac\"] = [\n",
    "    compute_percent(moa_protein_match[1]),\n",
    "    0,\n",
    "    compute_percent(moa_protein_match[1]),\n",
    "]\n",
    "summary_df_patient.loc[\"ClinVar\"] = [\n",
    "    0,\n",
    "    compute_percent(clinvar_genomic_match[1]),\n",
    "    compute_percent(clinvar_genomic_match[1]),\n",
    "]\n",
    "summary_df_patient.loc[\"CIViC\"] = [\n",
    "    compute_percent(civic_protein_match[1]),\n",
    "    compute_percent(civic_genomic_match[1]),\n",
    "    compute_percent(civic_match[1]),\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "1f94b1a8",
   "metadata": {},
   "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>Protein</th>\n",
       "      <th>Genomic</th>\n",
       "      <th>Protein or Genomic</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>All Knowledgebases</th>\n",
       "      <td>58.4</td>\n",
       "      <td>86.6</td>\n",
       "      <td>87.4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>MOAlmanac</th>\n",
       "      <td>28.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>28.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ClinVar</th>\n",
       "      <td>0.0</td>\n",
       "      <td>86.5</td>\n",
       "      <td>86.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CIViC</th>\n",
       "      <td>57.8</td>\n",
       "      <td>0.6</td>\n",
       "      <td>58.2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                    Protein  Genomic  Protein or Genomic\n",
       "All Knowledgebases     58.4     86.6                87.4\n",
       "MOAlmanac              28.5      0.0                28.5\n",
       "ClinVar                 0.0     86.5                86.5\n",
       "CIViC                  57.8      0.6                58.2"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "summary_df_patient"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "5e364633",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 900x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Create Grouped Bar Plot for Table 2\n",
    "summary_df_patient_plot = summary_df_patient.T\n",
    "summary_df_patient_plot = summary_df_patient_plot[\n",
    "    [\"MOAlmanac\", \"CIViC\", \"ClinVar\", \"All Knowledgebases\"]\n",
    "]\n",
    "patient_plot = summary_df_patient_plot.plot.bar(\n",
    "    rot=0, width=0.7, color=[\"#99DDFF\", \"#44BB99\", \"#FFAABB\", \"#DDDDDD\"], figsize=(9, 6)\n",
    ")\n",
    "for container in patient_plot.containers:\n",
    "    patient_plot.bar_label(container)\n",
    "patient_plot.legend(loc=\"upper right\", ncols=4, fontsize=12, prop={\"family\": \"Arial\"})\n",
    "patient_plot.set_ylim(0, 110)\n",
    "patient_plot.set_ylabel(\"Percent of Patients with Match\", fontsize=16, fontname=\"Arial\")\n",
    "patient_plot.set_xlabel(\"Variant Type\", fontsize=16, fontname=\"Arial\")\n",
    "patient_plot.tick_params(axis=\"x\", labelsize=14, labelfontfamily=\"Arial\")\n",
    "patient_plot.set_title(\n",
    "    \"Percent of GENIE Patients with At Least One\\nVariant Match to a Knowledgebase\",\n",
    "    fontsize=18,\n",
    "    fontname=\"Arial\",\n",
    "    weight=\"bold\",\n",
    ")\n",
    "patient_plot.figure.savefig(\"genie_patient_matching.png\", bbox_inches=\"tight\", dpi=1000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "e84ccb12",
   "metadata": {},
   "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>Protein</th>\n",
       "      <th>Genomic</th>\n",
       "      <th>Protein or Genomic</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>All Knowledgebases</th>\n",
       "      <td>58.4</td>\n",
       "      <td>86.6</td>\n",
       "      <td>87.4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>MOAlmanac</th>\n",
       "      <td>28.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>28.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ClinVar</th>\n",
       "      <td>0.0</td>\n",
       "      <td>86.5</td>\n",
       "      <td>86.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CIViC</th>\n",
       "      <td>57.8</td>\n",
       "      <td>0.6</td>\n",
       "      <td>58.2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                    Protein  Genomic  Protein or Genomic\n",
       "All Knowledgebases     58.4     86.6                87.4\n",
       "MOAlmanac              28.5      0.0                28.5\n",
       "ClinVar                 0.0     86.5                86.5\n",
       "CIViC                  57.8      0.6                58.2"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "summary_df_patient"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "fd602561",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Table 3 - Average Variants Per Patient. This average is calculated using the number of patients from each filtering category.\n",
    "summary_df_avg = pd.DataFrame(columns=[\"Protein\", \"Genomic\"])\n",
    "summary_df_avg.loc[\"Average # of Total Variants per Patient\"] = [\n",
    "    get_kb_avg_vars_per_patient(\n",
    "        genie_variants_df[genie_variants_df[\"free_text_p_short\"].notna()]\n",
    "    )[0],\n",
    "    get_kb_avg_vars_per_patient(\n",
    "        genie_variants_df[genie_variants_df[\"coordinates\"].notna()]\n",
    "    )[0],\n",
    "]\n",
    "summary_df_avg.loc[\"Average # of Normalized Variants per Patient\"] = [\n",
    "    get_kb_avg_vars_per_patient(\n",
    "        normalized_variants_df[normalized_variants_df[\"vrs_id_protein\"].notna()]\n",
    "    )[0],\n",
    "    get_kb_avg_vars_per_patient(\n",
    "        normalized_variants_df[normalized_variants_df[\"vrs_id_genomic\"].notna()]\n",
    "    )[0],\n",
    "]\n",
    "summary_df_avg.loc[\"Average # of Normalized Variants per Patient in Clinvar\"] = [\n",
    "    0,\n",
    "    clinvar_genomic_match[0],\n",
    "]\n",
    "summary_df_avg.loc[\"Average # of Normalized Variants per Patient in MOA\"] = [\n",
    "    moa_protein_match[0],\n",
    "    0,\n",
    "]\n",
    "summary_df_avg.loc[\"Average # of Normalized Variants per Patient in CIViC\"] = [\n",
    "    civic_protein_match[0],\n",
    "    civic_genomic_match[0],\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "200f687a",
   "metadata": {},
   "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>Protein</th>\n",
       "      <th>Genomic</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Average # of Total Variants per Patient</th>\n",
       "      <td>9.60</td>\n",
       "      <td>8.71</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Average # of Normalized Variants per Patient</th>\n",
       "      <td>7.96</td>\n",
       "      <td>8.70</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Average # of Normalized Variants per Patient in Clinvar</th>\n",
       "      <td>0.00</td>\n",
       "      <td>3.22</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Average # of Normalized Variants per Patient in MOA</th>\n",
       "      <td>1.10</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Average # of Normalized Variants per Patient in CIViC</th>\n",
       "      <td>1.38</td>\n",
       "      <td>1.01</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                    Protein  Genomic\n",
       "Average # of Total Variants per Patient                9.60     8.71\n",
       "Average # of Normalized Variants per Patient           7.96     8.70\n",
       "Average # of Normalized Variants per Patient in...     0.00     3.22\n",
       "Average # of Normalized Variants per Patient in...     1.10     0.00\n",
       "Average # of Normalized Variants per Patient in...     1.38     1.01"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "summary_df_avg"
   ]
  }
 ],
 "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": "3.11.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}