2021-BUSCO-Gene-Selection.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "import glob\n",
    "import os\n",
    "import pandas as pd\n",
    "import matplotlib as mpl\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from datetime import date\n",
    "from Bio.SeqRecord import SeqRecord\n",
    "from Bio.Seq import Seq\n",
    "from Bio import SeqIO\n",
    "import numpy as np\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "# SET PATHS FOR GENERAL BUSCO LOCATION\n",
    "BUSCO_PATHS = '/vortexfs1/omics/alexander/halexander/2020-05-euk-abund/phylogenomics/reference-genomes/busco_downloads/lineages/eukaryota_odb10/info/ogs.id.info'\n",
    "BUSCO_LIST = list(set(list(pd.read_csv(BUSCO_PATHS, sep='\\t', header=None)[1])))\n",
    "\n",
    "#BASE PATH TO MAG BUSCO RUNS\n",
    "\n",
    "# TOPAZ MAGs\n",
    "MAG_BASE_PATH=os.path.join('/vortexfs1/scratch/halexander/2020-eukms-annotate/')\n",
    "MAG_DIR = os.path.join(MAG_BASE_PATH, 'busco/eukaryota_odb10')\n",
    "MAG_BASE_TSV = pd.read_csv(os.path.join(MAG_BASE_PATH, 'busco-update-2021-01-12.tsv'), \n",
    "            sep='\\t', index_col=0)\n",
    "MAG_BASE_TSV['Proportion_Complete'] = (MAG_BASE_TSV.Complete_ALL + MAG_BASE_TSV.Fragmented)/255\n",
    "\n",
    "# DELMONT MAGs\n",
    "DMAG_BASE_PATH=os.path.join('/vortexfs1/omics/alexander/halexander/2020-delmont')\n",
    "DMAG_DIR = os.path.join(DMAG_BASE_PATH, 'busco/eukaryota_odb10')\n",
    "DMAG_BASE_TSV = pd.read_csv(os.path.join(DMAG_BASE_PATH, 'busco-summary.tsv'), \n",
    "            sep='\\t', index_col=0)\n",
    "DMAG_BASE_TSV['Proportion_Complete'] = (DMAG_BASE_TSV.Complete_ALL + DMAG_BASE_TSV.Fragmented)/255\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "# SET PROPORTION CUTOFF TO INCLUDE\n",
    "BUSCO_COMPLETENESS_CUTOFF = .5\n",
    "\n",
    "MAG_COMPLETE_LIST = MAG_BASE_TSV.loc[MAG_BASE_TSV.Proportion_Complete>=BUSCO_COMPLETENESS_CUTOFF].index\n",
    "DMAG_COMPLETE_LIST = DMAG_BASE_TSV.loc[DMAG_BASE_TSV.Proportion_Complete>=BUSCO_COMPLETENESS_CUTOFF].index\n",
    "            \n",
    "                       "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Loop through identified completed MAGs (>50% BUSCO) and create a table that contains a boolean related to the presence / absence of the protein within the MAG"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [],
   "source": [
    "def GetBUSCOBoolean(MAG_DIR, MAG_COMPLETE_LIST, BUSCO_LIST):\n",
    "#     i=MAG_COMPLETE_LIST[0]\n",
    "    COMPLETE_DF = pd.DataFrame(columns = MAG_COMPLETE_LIST,index=BUSCO_LIST)\n",
    "    COMPLETE_DF=COMPLETE_DF.fillna(0)\n",
    "    COMPLETE_FRAGMENTED_DF=COMPLETE_DF.fillna(0)\n",
    "    for i in MAG_COMPLETE_LIST:\n",
    "        df=pd.read_csv(os.path.join(MAG_DIR, i, 'run_eukaryota_odb10', \n",
    "                'full_table.tsv'), sep='\\t', header=2, index_col=0)\n",
    "        complete=df.loc[df.Status == 'Complete'].index\n",
    "        fragmented=df.loc[df.Status == 'Fragmented'].index\n",
    "        COMPLETE_DF.loc[complete,i]=1\n",
    "        COMPLETE_FRAGMENTED_DF.loc[fragmented,i]=1\n",
    "        COMPLETE_FRAGMENTED_DF.loc[complete,i]=1\n",
    "    return(COMPLETE_DF, COMPLETE_FRAGMENTED_DF)\n",
    "\n",
    "MAG_C_BOO,MAG_CF_BOO = GetBUSCOBoolean(MAG_DIR, MAG_COMPLETE_LIST, BUSCO_LIST)\n",
    "DMAG_C_BOO,DMAG_CF_BOO = GetBUSCOBoolean(DMAG_DIR, DMAG_COMPLETE_LIST, BUSCO_LIST)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 87,
     "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": [
    "# Plot of the relative coverage of protins across buscos\n",
    "(MAG_CF_BOO.sum(axis=1)/len(MAG_CF_BOO.columns)).hist(bins=20)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 88,
     "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": [
    "(DMAG_CF_BOO.sum(axis=1)/len(DMAG_CF_BOO.columns)).hist(bins=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [],
   "source": [
    "# BUSCO cutoff\n",
    "\n",
    "COV90 = MAG_CF_BOO.loc[(MAG_CF_BOO.sum(axis=1)/len(MAG_CF_BOO.columns))>=.90].index\n",
    "COV80 = MAG_CF_BOO.loc[(MAG_CF_BOO.sum(axis=1)/len(MAG_CF_BOO.columns))>=.80].index\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 114,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "## Check coverage in our BUSCO set of MAGs\n",
    "fig, ax =plt.subplots(2)\n",
    "(MAG_CF_BOO.loc[COV90].sum()/len(COV90)).hist(bins=10, range=(0,1), ax=ax[0])\n",
    "(DMAG_CF_BOO.loc[COV90].sum()/len(COV90)).hist(bins=10, range=(0,1), ax = ax[1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Reference genomes\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Reference Genomes BUSCO hits\n",
    "REF_DIR='/vortexfs1/omics/alexander/halexander/2020-05-euk-abund/phylogenomics/reference-genomes/busco/eukaryota_odb10'\n",
    "\n",
    "REF_GENOMES=[]\n",
    "for o in glob.glob(os.path.join(REF_DIR,'*')):\n",
    "    REF_GENOMES.append(os.path.basename(o))\n",
    "\n",
    "REF_busco_completeness_df = pd.DataFrame(columns = REF_GENOMES,index=BUSCO_LIST)\n",
    "REF_busco_completeness_df=REF_busco_completeness_df.fillna(0)\n",
    "REF_busco_completeness_fragmented_df=REF_busco_completeness_df.fillna(0)\n",
    "for i in REF_GENOMES:\n",
    "    df=pd.read_csv(os.path.join(REF_DIR, i, 'run_eukaryota_odb10', \n",
    "                'full_table.tsv'), sep='\\t', header=2, index_col=0)\n",
    "    complete=df.loc[df.Status == 'Complete'].index\n",
    "    fragmented=df.loc[df.Status == 'Fragmented'].index\n",
    "    REF_busco_completeness_df.loc[complete,i]=1\n",
    "    REF_busco_completeness_fragmented_df.loc[fragmented,i]=1\n",
    "    REF_busco_completeness_fragmented_df.loc[complete,i]=1\n",
    "\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 125,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax =plt.subplots(2)\n",
    "\n",
    "(REF_busco_completeness_fragmented_df.loc[COV90].sum()/len(COV90)).hist(bins=10, range=(0,1), ax=ax[0])\n",
    "(REF_busco_completeness_fragmented_df.loc[COV80].sum()/len(COV80)).hist(bins=10, range=(0,1), ax=ax[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['777920at2759', '341721at2759', '1106766at2759', '1423847at2759',\n",
       "       '1014314at2759', '1197019at2759', '625387at2759', '551907at2759',\n",
       "       '1259741at2759', '1479417at2759', '369837at2759', '1346432at2759',\n",
       "       '1025450at2759', '426305at2759', '1545004at2759', '1355894at2759',\n",
       "       '917326at2759', '570797at2759', '664730at2759', '1010730at2759',\n",
       "       '1287094at2759', '1379600at2759', '166920at2759', '834694at2759',\n",
       "       '905026at2759', '257318at2759', '83779at2759', '330169at2759',\n",
       "       '1538526at2759', '975557at2759', '604979at2759', '100698at2759',\n",
       "       '390348at2759', '299766at2759', '418107at2759', '734666at2759',\n",
       "       '836599at2759', '720340at2759', '1248958at2759', '375960at2759',\n",
       "       '931188at2759', '1249159at2759', '1182451at2759', '179362at2759',\n",
       "       '97116at2759', '549762at2759', '1217666at2759', '355408at2759',\n",
       "       '774318at2759', '736068at2759', '513979at2759', '1194797at2759',\n",
       "       '1260807at2759', '1085752at2759', '296129at2759', '761109at2759',\n",
       "       '1314980at2759'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 126,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "proteins"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['RS-all-SRF-0-8-5-00_bin-105', 'SPO-all-SRF-0-8-5-00_bin-190',\n",
       "       'NAO-all-MES-0-8-5-00_bin-73', 'SAO-all-SRF-20-180-00_bin-42',\n",
       "       'NAO-all-SRF-0-8-5-00_bin-17', 'SPO-SPSG-MIX-0-8-5-00_bin-132',\n",
       "       'MS-all-DCM-0-8-5-00_bin-224', 'MS-all-DCM-0-8-5-00_bin-636',\n",
       "       'SPO-all-SRF-5-20-00_bin-108', 'SPO-SPSG-DCM-0-8-5-00_bin-172',\n",
       "       ...\n",
       "       'MS-all-SRF-180-2000-00_bin-83', 'SAO-all-SRF-0-8-5-00_bin-26',\n",
       "       'SPO-SPSG-MES-0-8-5-00_bin-179', 'SAO-all-DCM-0-8-5-00_bin-428',\n",
       "       'SPO-SPSG-SRF-0-8-5-00_bin-658', 'SPO-SPSG-SRF-0-8-5-00_bin-277',\n",
       "       'NAO-all-DCM-0-8-5-00_bin-77', 'SAO-all-DCM-0-8-5-00_bin-2',\n",
       "       'SPO-CHIL-DCM-0-8-5-00_bin-46', 'IO-all-SRF-0-8-5-00_bin-89'],\n",
       "      dtype='object', length=336)"
      ]
     },
     "execution_count": 127,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "MAGs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 149,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "def WriteConcatenatedProteins(proteins, MAG_DIR, MAGs, outdir=os.path.join('extracted', str(date.today()))):\n",
    "    if not os.path.exists(outdir):\n",
    "            os.makedirs(outdir)\n",
    "    for j, o in enumerate(proteins):\n",
    "        all_og=[]\n",
    "        og_file = os.path.join(outdir,o+'.concatenated.faa')\n",
    "        for i, m in enumerate(MAGs):\n",
    "            og_path=''\n",
    "            complete_path = os.path.join(MAG_DIR, m, \n",
    "                    'run_eukaryota_odb10','busco_sequences',\n",
    "                    'single_copy_busco_sequences', o+'.faa')\n",
    "            frag_path = os.path.join(MAG_DIR, m, \n",
    "                    'run_eukaryota_odb10','busco_sequences',\n",
    "                    'fragmented_busco_sequences', o+'.faa')\n",
    "            #check for complete OG first\n",
    "            if os.path.isfile(complete_path):\n",
    "                og_path = complete_path\n",
    "            #check for fragmented OG\n",
    "            elif os.path.isfile(frag_path):\n",
    "                og_path = frag_path\n",
    "            #Otherwise just skip it... \n",
    "            if og_path!='':\n",
    "                for record in SeqIO.parse(og_path, \"fasta\"):\n",
    "                    record.id = m\n",
    "                    record.name=m\n",
    "                    all_og.append(record)\n",
    "            #create an empty sequence record with an 'X'-- undetermined amino acid. \n",
    "            else: \n",
    "                record = SeqRecord(Seq(\"-\"),id=m,name=m,description=\"no-record\")\n",
    "                all_og.append(record)\n",
    "#     for i, m in enumerate(REF_GENOMES):\n",
    "#         og_path=''\n",
    "#         complete_path = os.path.join(REF_DIR, m, \n",
    "#                 'run_eukaryota_odb10','busco_sequences',\n",
    "#                 'single_copy_busco_sequences', o+'.faa')\n",
    "#         frag_path = os.path.join(REF_DIR, m, \n",
    "#                 'run_eukaryota_odb10','busco_sequences',\n",
    "#                 'fragmented_busco_sequences', o+'.faa')\n",
    "#         #check for complete OG first\n",
    "#         if os.path.isfile(complete_path):\n",
    "#             og_path = complete_path\n",
    "#         #check for fragmented OG\n",
    "#         elif os.path.isfile(frag_path):\n",
    "#             og_path = frag_path\n",
    "#         #Otherwise just skip it... \n",
    "#         if og_path!='':\n",
    "#             for record in SeqIO.parse(og_path, \"fasta\"):\n",
    "#                 record.id = m\n",
    "#                 record.name=m\n",
    "#                 all_og.append(record)\n",
    "#         #create an empty sequence record with an 'X'-- undetermined amino acid. \n",
    "#         else: \n",
    "#             record = SeqRecord(Seq(\"-\"),id=m,name=m,description=\"no-record\")\n",
    "#             all_og.append(record)\n",
    "        if os.path.isfile(og_file):\n",
    "            with open(og_file, \"a\") as outfasta:\n",
    "                SeqIO.write(all_og, outfasta,'fasta')\n",
    "            outfasta.close()    \n",
    "        else:\n",
    "            with open(og_file, \"w\") as outfasta:\n",
    "                SeqIO.write(all_og, outfasta,'fasta')\n",
    "            outfasta.close()    \n",
    "    if os.path.isfile(os.path.join(outdir,'ALLMAGS.list')):\n",
    "        with open(os.path.join(outdir,'ALLMAGS.list'), 'a') as f:\n",
    "            f.write('\\n'.join(MAGs))\n",
    "            f.write('\\n')\n",
    "    else: \n",
    "        with open(os.path.join(outdir,'ALLMAGS.list'), 'w') as f:\n",
    "            f.write('\\n'.join(MAGs))\n",
    "            f.write('\\n')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 151,
   "metadata": {},
   "outputs": [],
   "source": [
    "outdir = os.path.join('extracted','COV90'+str(date.today()))\n",
    "WriteConcatenatedProteins(COV90, MAG_DIR, MAG_COMPLETE_LIST, outdir=outdir)\n",
    "WriteConcatenatedProteins(COV90, DMAG_DIR, DMAG_COMPLETE_LIST, outdir=outdir)\n",
    "WriteConcatenatedProteins(COV90, REF_DIR, REF_GENOMES, outdir=outdir)\n",
    "with open(os.path.join(outdir,'ALLPROTEINS.list'), 'w') as g:\n",
    "    g.write('\\n'.join(COV90))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 154,
   "metadata": {},
   "outputs": [],
   "source": [
    "outdir = os.path.join('extracted','COV80'+str(date.today()))\n",
    "WriteConcatenatedProteins(COV80, MAG_DIR, MAG_COMPLETE_LIST, outdir=outdir)\n",
    "WriteConcatenatedProteins(COV80, DMAG_DIR, DMAG_COMPLETE_LIST, outdir=outdir)\n",
    "WriteConcatenatedProteins(COV80, REF_DIR, REF_GENOMES, outdir=outdir)\n",
    "with open(os.path.join(outdir,'ALLPROTEINS.list'), 'w') as g:\n",
    "    g.write('\\n'.join(proteins))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "10"
      ]
     },
     "execution_count": 159,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "REF_busco_completeness_fragmented_df.loc[COV90,'MMETSP0818'].sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 162,
   "metadata": {},
   "outputs": [],
   "source": [
    "outdir = os.path.join('extracted','COV90_TOPAZ_'+str(date.today()))\n",
    "WriteConcatenatedProteins(COV90, MAG_DIR, MAG_COMPLETE_LIST, outdir=outdir)\n",
    "with open(os.path.join(outdir,'ALLPROTEINS.list'), 'w') as g:\n",
    "    g.write('\\n'.join(COV90))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "metadata": {},
   "outputs": [],
   "source": [
    "outdir = os.path.join('extracted','COV80_TOPAZ_'+str(date.today()))\n",
    "WriteConcatenatedProteins(COV80, MAG_DIR, MAG_COMPLETE_LIST, outdir=outdir)\n",
    "with open(os.path.join(outdir,'ALLPROTEINS.list'), 'w') as g:\n",
    "    g.write('\\n'.join(proteins))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<function __main__.readin_ani(file)>"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "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>1</th>\n",
       "      <th>IO-all-DCM-0-8-5-00_bin-104</th>\n",
       "      <th>IO-all-DCM-0-8-5-00_bin-107</th>\n",
       "      <th>IO-all-DCM-0-8-5-00_bin-111</th>\n",
       "      <th>IO-all-DCM-0-8-5-00_bin-12</th>\n",
       "      <th>IO-all-DCM-0-8-5-00_bin-120</th>\n",
       "      <th>IO-all-DCM-0-8-5-00_bin-130</th>\n",
       "      <th>IO-all-DCM-0-8-5-00_bin-147</th>\n",
       "      <th>IO-all-DCM-0-8-5-00_bin-157</th>\n",
       "      <th>IO-all-DCM-0-8-5-00_bin-17</th>\n",
       "      <th>IO-all-DCM-0-8-5-00_bin-21</th>\n",
       "      <th>...</th>\n",
       "      <th>TARA_SOC_28_MAG_00071</th>\n",
       "      <th>TARA_SOC_28_MAG_00072</th>\n",
       "      <th>TARA_SOC_28_MAG_00073</th>\n",
       "      <th>TARA_SOC_28_MAG_00074</th>\n",
       "      <th>TARA_SOC_28_MAG_00075</th>\n",
       "      <th>TARA_SOC_28_MAG_00076</th>\n",
       "      <th>TARA_SOC_28_MAG_00077</th>\n",
       "      <th>TARA_SOC_28_MAG_00078</th>\n",
       "      <th>TARA_SOC_28_MAG_00079</th>\n",
       "      <th>TARA_SOC_28_MAG_00080</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</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>IO-all-DCM-0-8-5-00_bin-104</th>\n",
       "      <td>100.0000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.5657</td>\n",
       "      <td>74.5590</td>\n",
       "      <td>74.6903</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.5737</td>\n",
       "      <td>74.6215</td>\n",
       "      <td>74.5590</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.7430</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.6856</td>\n",
       "      <td>74.6613</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.6759</td>\n",
       "      <td>74.5949</td>\n",
       "      <td>74.6262</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>IO-all-DCM-0-8-5-00_bin-107</th>\n",
       "      <td>NaN</td>\n",
       "      <td>100.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>IO-all-DCM-0-8-5-00_bin-111</th>\n",
       "      <td>74.5898</td>\n",
       "      <td>NaN</td>\n",
       "      <td>100.0000</td>\n",
       "      <td>75.1033</td>\n",
       "      <td>74.7017</td>\n",
       "      <td>74.7280</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.7226</td>\n",
       "      <td>74.5991</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.6241</td>\n",
       "      <td>NaN</td>\n",
       "      <td>75.1093</td>\n",
       "      <td>74.7165</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.5724</td>\n",
       "      <td>74.7943</td>\n",
       "      <td>74.6783</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>IO-all-DCM-0-8-5-00_bin-12</th>\n",
       "      <td>74.5837</td>\n",
       "      <td>NaN</td>\n",
       "      <td>75.1133</td>\n",
       "      <td>100.0000</td>\n",
       "      <td>74.6443</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.6069</td>\n",
       "      <td>74.5550</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.6296</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.9889</td>\n",
       "      <td>74.6768</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.5963</td>\n",
       "      <td>74.7040</td>\n",
       "      <td>74.6473</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>IO-all-DCM-0-8-5-00_bin-120</th>\n",
       "      <td>74.6746</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.6935</td>\n",
       "      <td>74.6479</td>\n",
       "      <td>100.0000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.8074</td>\n",
       "      <td>74.6004</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.6902</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.7656</td>\n",
       "      <td>74.7031</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.7000</td>\n",
       "      <td>74.7415</td>\n",
       "      <td>74.7449</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TARA_SOC_28_MAG_00076</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.5711</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.5617</td>\n",
       "      <td>74.8519</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.5606</td>\n",
       "      <td>100.0000</td>\n",
       "      <td>74.5643</td>\n",
       "      <td>74.5679</td>\n",
       "      <td>74.5248</td>\n",
       "      <td>75.6425</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TARA_SOC_28_MAG_00077</th>\n",
       "      <td>74.7022</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.5940</td>\n",
       "      <td>74.6236</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.6426</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.5948</td>\n",
       "      <td>...</td>\n",
       "      <td>74.7463</td>\n",
       "      <td>74.6819</td>\n",
       "      <td>74.5606</td>\n",
       "      <td>74.6994</td>\n",
       "      <td>74.6473</td>\n",
       "      <td>74.5726</td>\n",
       "      <td>100.0000</td>\n",
       "      <td>74.7196</td>\n",
       "      <td>74.6303</td>\n",
       "      <td>74.6932</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TARA_SOC_28_MAG_00078</th>\n",
       "      <td>74.6586</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.7244</td>\n",
       "      <td>74.6686</td>\n",
       "      <td>74.6599</td>\n",
       "      <td>74.6430</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.6829</td>\n",
       "      <td>74.7074</td>\n",
       "      <td>74.6844</td>\n",
       "      <td>...</td>\n",
       "      <td>74.6935</td>\n",
       "      <td>75.1744</td>\n",
       "      <td>74.6426</td>\n",
       "      <td>75.0133</td>\n",
       "      <td>75.0503</td>\n",
       "      <td>74.6413</td>\n",
       "      <td>74.7938</td>\n",
       "      <td>100.0000</td>\n",
       "      <td>75.1778</td>\n",
       "      <td>74.6241</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TARA_SOC_28_MAG_00079</th>\n",
       "      <td>74.6415</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.6682</td>\n",
       "      <td>74.6505</td>\n",
       "      <td>74.7616</td>\n",
       "      <td>74.7534</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.6169</td>\n",
       "      <td>74.6818</td>\n",
       "      <td>74.5906</td>\n",
       "      <td>...</td>\n",
       "      <td>74.6696</td>\n",
       "      <td>75.2482</td>\n",
       "      <td>74.5843</td>\n",
       "      <td>74.7807</td>\n",
       "      <td>74.9443</td>\n",
       "      <td>74.6012</td>\n",
       "      <td>74.6382</td>\n",
       "      <td>75.2315</td>\n",
       "      <td>100.0000</td>\n",
       "      <td>74.7075</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TARA_SOC_28_MAG_00080</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.4924</td>\n",
       "      <td>78.2689</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>75.7934</td>\n",
       "      <td>74.6518</td>\n",
       "      <td>74.6348</td>\n",
       "      <td>74.7172</td>\n",
       "      <td>100.0000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1708 rows × 1708 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "1                            IO-all-DCM-0-8-5-00_bin-104  \\\n",
       "0                                                          \n",
       "IO-all-DCM-0-8-5-00_bin-104                     100.0000   \n",
       "IO-all-DCM-0-8-5-00_bin-107                          NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-111                      74.5898   \n",
       "IO-all-DCM-0-8-5-00_bin-12                       74.5837   \n",
       "IO-all-DCM-0-8-5-00_bin-120                      74.6746   \n",
       "...                                                  ...   \n",
       "TARA_SOC_28_MAG_00076                                NaN   \n",
       "TARA_SOC_28_MAG_00077                            74.7022   \n",
       "TARA_SOC_28_MAG_00078                            74.6586   \n",
       "TARA_SOC_28_MAG_00079                            74.6415   \n",
       "TARA_SOC_28_MAG_00080                                NaN   \n",
       "\n",
       "1                            IO-all-DCM-0-8-5-00_bin-107  \\\n",
       "0                                                          \n",
       "IO-all-DCM-0-8-5-00_bin-104                          NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-107                        100.0   \n",
       "IO-all-DCM-0-8-5-00_bin-111                          NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-12                           NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-120                          NaN   \n",
       "...                                                  ...   \n",
       "TARA_SOC_28_MAG_00076                                NaN   \n",
       "TARA_SOC_28_MAG_00077                                NaN   \n",
       "TARA_SOC_28_MAG_00078                                NaN   \n",
       "TARA_SOC_28_MAG_00079                                NaN   \n",
       "TARA_SOC_28_MAG_00080                                NaN   \n",
       "\n",
       "1                            IO-all-DCM-0-8-5-00_bin-111  \\\n",
       "0                                                          \n",
       "IO-all-DCM-0-8-5-00_bin-104                      74.5657   \n",
       "IO-all-DCM-0-8-5-00_bin-107                          NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-111                     100.0000   \n",
       "IO-all-DCM-0-8-5-00_bin-12                       75.1133   \n",
       "IO-all-DCM-0-8-5-00_bin-120                      74.6935   \n",
       "...                                                  ...   \n",
       "TARA_SOC_28_MAG_00076                                NaN   \n",
       "TARA_SOC_28_MAG_00077                            74.5940   \n",
       "TARA_SOC_28_MAG_00078                            74.7244   \n",
       "TARA_SOC_28_MAG_00079                            74.6682   \n",
       "TARA_SOC_28_MAG_00080                                NaN   \n",
       "\n",
       "1                            IO-all-DCM-0-8-5-00_bin-12  \\\n",
       "0                                                         \n",
       "IO-all-DCM-0-8-5-00_bin-104                     74.5590   \n",
       "IO-all-DCM-0-8-5-00_bin-107                         NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-111                     75.1033   \n",
       "IO-all-DCM-0-8-5-00_bin-12                     100.0000   \n",
       "IO-all-DCM-0-8-5-00_bin-120                     74.6479   \n",
       "...                                                 ...   \n",
       "TARA_SOC_28_MAG_00076                               NaN   \n",
       "TARA_SOC_28_MAG_00077                           74.6236   \n",
       "TARA_SOC_28_MAG_00078                           74.6686   \n",
       "TARA_SOC_28_MAG_00079                           74.6505   \n",
       "TARA_SOC_28_MAG_00080                               NaN   \n",
       "\n",
       "1                            IO-all-DCM-0-8-5-00_bin-120  \\\n",
       "0                                                          \n",
       "IO-all-DCM-0-8-5-00_bin-104                      74.6903   \n",
       "IO-all-DCM-0-8-5-00_bin-107                          NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-111                      74.7017   \n",
       "IO-all-DCM-0-8-5-00_bin-12                       74.6443   \n",
       "IO-all-DCM-0-8-5-00_bin-120                     100.0000   \n",
       "...                                                  ...   \n",
       "TARA_SOC_28_MAG_00076                                NaN   \n",
       "TARA_SOC_28_MAG_00077                                NaN   \n",
       "TARA_SOC_28_MAG_00078                            74.6599   \n",
       "TARA_SOC_28_MAG_00079                            74.7616   \n",
       "TARA_SOC_28_MAG_00080                                NaN   \n",
       "\n",
       "1                            IO-all-DCM-0-8-5-00_bin-130  \\\n",
       "0                                                          \n",
       "IO-all-DCM-0-8-5-00_bin-104                          NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-107                          NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-111                      74.7280   \n",
       "IO-all-DCM-0-8-5-00_bin-12                           NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-120                          NaN   \n",
       "...                                                  ...   \n",
       "TARA_SOC_28_MAG_00076                                NaN   \n",
       "TARA_SOC_28_MAG_00077                                NaN   \n",
       "TARA_SOC_28_MAG_00078                            74.6430   \n",
       "TARA_SOC_28_MAG_00079                            74.7534   \n",
       "TARA_SOC_28_MAG_00080                                NaN   \n",
       "\n",
       "1                            IO-all-DCM-0-8-5-00_bin-147  \\\n",
       "0                                                          \n",
       "IO-all-DCM-0-8-5-00_bin-104                          NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-107                          NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-111                          NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-12                           NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-120                          NaN   \n",
       "...                                                  ...   \n",
       "TARA_SOC_28_MAG_00076                                NaN   \n",
       "TARA_SOC_28_MAG_00077                                NaN   \n",
       "TARA_SOC_28_MAG_00078                                NaN   \n",
       "TARA_SOC_28_MAG_00079                                NaN   \n",
       "TARA_SOC_28_MAG_00080                                NaN   \n",
       "\n",
       "1                            IO-all-DCM-0-8-5-00_bin-157  \\\n",
       "0                                                          \n",
       "IO-all-DCM-0-8-5-00_bin-104                      74.5737   \n",
       "IO-all-DCM-0-8-5-00_bin-107                          NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-111                          NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-12                           NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-120                          NaN   \n",
       "...                                                  ...   \n",
       "TARA_SOC_28_MAG_00076                            74.5711   \n",
       "TARA_SOC_28_MAG_00077                            74.6426   \n",
       "TARA_SOC_28_MAG_00078                            74.6829   \n",
       "TARA_SOC_28_MAG_00079                            74.6169   \n",
       "TARA_SOC_28_MAG_00080                                NaN   \n",
       "\n",
       "1                            IO-all-DCM-0-8-5-00_bin-17  \\\n",
       "0                                                         \n",
       "IO-all-DCM-0-8-5-00_bin-104                     74.6215   \n",
       "IO-all-DCM-0-8-5-00_bin-107                         NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-111                     74.7226   \n",
       "IO-all-DCM-0-8-5-00_bin-12                      74.6069   \n",
       "IO-all-DCM-0-8-5-00_bin-120                     74.8074   \n",
       "...                                                 ...   \n",
       "TARA_SOC_28_MAG_00076                               NaN   \n",
       "TARA_SOC_28_MAG_00077                               NaN   \n",
       "TARA_SOC_28_MAG_00078                           74.7074   \n",
       "TARA_SOC_28_MAG_00079                           74.6818   \n",
       "TARA_SOC_28_MAG_00080                               NaN   \n",
       "\n",
       "1                            IO-all-DCM-0-8-5-00_bin-21  ...  \\\n",
       "0                                                        ...   \n",
       "IO-all-DCM-0-8-5-00_bin-104                     74.5590  ...   \n",
       "IO-all-DCM-0-8-5-00_bin-107                         NaN  ...   \n",
       "IO-all-DCM-0-8-5-00_bin-111                     74.5991  ...   \n",
       "IO-all-DCM-0-8-5-00_bin-12                      74.5550  ...   \n",
       "IO-all-DCM-0-8-5-00_bin-120                     74.6004  ...   \n",
       "...                                                 ...  ...   \n",
       "TARA_SOC_28_MAG_00076                               NaN  ...   \n",
       "TARA_SOC_28_MAG_00077                           74.5948  ...   \n",
       "TARA_SOC_28_MAG_00078                           74.6844  ...   \n",
       "TARA_SOC_28_MAG_00079                           74.5906  ...   \n",
       "TARA_SOC_28_MAG_00080                               NaN  ...   \n",
       "\n",
       "1                            TARA_SOC_28_MAG_00071  TARA_SOC_28_MAG_00072  \\\n",
       "0                                                                           \n",
       "IO-all-DCM-0-8-5-00_bin-104                    NaN                74.7430   \n",
       "IO-all-DCM-0-8-5-00_bin-107                    NaN                    NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-111                    NaN                74.6241   \n",
       "IO-all-DCM-0-8-5-00_bin-12                     NaN                74.6296   \n",
       "IO-all-DCM-0-8-5-00_bin-120                    NaN                74.6902   \n",
       "...                                            ...                    ...   \n",
       "TARA_SOC_28_MAG_00076                          NaN                74.5617   \n",
       "TARA_SOC_28_MAG_00077                      74.7463                74.6819   \n",
       "TARA_SOC_28_MAG_00078                      74.6935                75.1744   \n",
       "TARA_SOC_28_MAG_00079                      74.6696                75.2482   \n",
       "TARA_SOC_28_MAG_00080                          NaN                74.4924   \n",
       "\n",
       "1                            TARA_SOC_28_MAG_00073  TARA_SOC_28_MAG_00074  \\\n",
       "0                                                                           \n",
       "IO-all-DCM-0-8-5-00_bin-104                    NaN                74.6856   \n",
       "IO-all-DCM-0-8-5-00_bin-107                    NaN                    NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-111                    NaN                75.1093   \n",
       "IO-all-DCM-0-8-5-00_bin-12                     NaN                74.9889   \n",
       "IO-all-DCM-0-8-5-00_bin-120                    NaN                74.7656   \n",
       "...                                            ...                    ...   \n",
       "TARA_SOC_28_MAG_00076                      74.8519                    NaN   \n",
       "TARA_SOC_28_MAG_00077                      74.5606                74.6994   \n",
       "TARA_SOC_28_MAG_00078                      74.6426                75.0133   \n",
       "TARA_SOC_28_MAG_00079                      74.5843                74.7807   \n",
       "TARA_SOC_28_MAG_00080                      78.2689                    NaN   \n",
       "\n",
       "1                            TARA_SOC_28_MAG_00075  TARA_SOC_28_MAG_00076  \\\n",
       "0                                                                           \n",
       "IO-all-DCM-0-8-5-00_bin-104                74.6613                    NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-107                    NaN                    NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-111                74.7165                    NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-12                 74.6768                    NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-120                74.7031                    NaN   \n",
       "...                                            ...                    ...   \n",
       "TARA_SOC_28_MAG_00076                      74.5606               100.0000   \n",
       "TARA_SOC_28_MAG_00077                      74.6473                74.5726   \n",
       "TARA_SOC_28_MAG_00078                      75.0503                74.6413   \n",
       "TARA_SOC_28_MAG_00079                      74.9443                74.6012   \n",
       "TARA_SOC_28_MAG_00080                          NaN                75.7934   \n",
       "\n",
       "1                            TARA_SOC_28_MAG_00077  TARA_SOC_28_MAG_00078  \\\n",
       "0                                                                           \n",
       "IO-all-DCM-0-8-5-00_bin-104                74.6759                74.5949   \n",
       "IO-all-DCM-0-8-5-00_bin-107                    NaN                    NaN   \n",
       "IO-all-DCM-0-8-5-00_bin-111                74.5724                74.7943   \n",
       "IO-all-DCM-0-8-5-00_bin-12                 74.5963                74.7040   \n",
       "IO-all-DCM-0-8-5-00_bin-120                74.7000                74.7415   \n",
       "...                                            ...                    ...   \n",
       "TARA_SOC_28_MAG_00076                      74.5643                74.5679   \n",
       "TARA_SOC_28_MAG_00077                     100.0000                74.7196   \n",
       "TARA_SOC_28_MAG_00078                      74.7938               100.0000   \n",
       "TARA_SOC_28_MAG_00079                      74.6382                75.2315   \n",
       "TARA_SOC_28_MAG_00080                      74.6518                74.6348   \n",
       "\n",
       "1                            TARA_SOC_28_MAG_00079  TARA_SOC_28_MAG_00080  \n",
       "0                                                                          \n",
       "IO-all-DCM-0-8-5-00_bin-104                74.6262                    NaN  \n",
       "IO-all-DCM-0-8-5-00_bin-107                    NaN                    NaN  \n",
       "IO-all-DCM-0-8-5-00_bin-111                74.6783                    NaN  \n",
       "IO-all-DCM-0-8-5-00_bin-12                 74.6473                    NaN  \n",
       "IO-all-DCM-0-8-5-00_bin-120                74.7449                    NaN  \n",
       "...                                            ...                    ...  \n",
       "TARA_SOC_28_MAG_00076                      74.5248                75.6425  \n",
       "TARA_SOC_28_MAG_00077                      74.6303                74.6932  \n",
       "TARA_SOC_28_MAG_00078                      75.1778                74.6241  \n",
       "TARA_SOC_28_MAG_00079                     100.0000                74.7075  \n",
       "TARA_SOC_28_MAG_00080                      74.7172               100.0000  \n",
       "\n",
       "[1708 rows x 1708 columns]"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fani_comp"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:.conda-general_plotting] *",
   "language": "python",
   "name": "conda-env-.conda-general_plotting-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": "3.6.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}