Adding files from scib

notebooks/analysis/ATAC_feature_analysis.ipynb

@@ -0,0 +1,271 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import pandas as pd\n",
+    "import numpy as np\n",
+    "from collections import defaultdict\n",
+    "import seaborn as sns\n",
+    "import matplotlib.pyplot as plt"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "atac1 = pd.read_csv('../../../Paper/202010_Revision/Supplementary Files/Results/ATAC/mouse_brain_atac_large_11batches_summary_scores.csv')\n",
+    "atac2 = pd.read_csv('../../../Paper/202010_Revision/Supplementary Files/Results/ATAC/mouse_brain_atac_small_summary_scores.csv')\n",
+    "atac = pd.concat([atac1,atac2])"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Average bio conservation score of methods for atac"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "Index(['Unnamed: 0', 'Method', 'Output', 'Feature Space', 'Overall Score',\n",
+       "       'Batch Correction', 'PCR batch', 'Batch ASW', 'iLISI',\n",
+       "       'graph connectivity', 'kBET', 'Bio conservation', 'NMI cluster/label',\n",
+       "       'ARI cluster/label', 'Cell type ASW', 'isolated label F1',\n",
+       "       'isolated label silhouette', 'cLISI'],\n",
+       "      dtype='object')"
+      ]
+     },
+     "execution_count": 3,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "atac1.columns"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Filter each dataset to only contain methods that are in all 3 feature spaces\n",
+    "\n",
+    "atac1 = atac1.loc[~np.isnan(atac1['Overall Score']),:]\n",
+    "atac1['meth_out'] = ['_'.join(atac1[['Method', 'Output']].values[i]) for i in range(atac1.shape[0])]\n",
+    "\n",
+    "meth_id_list = atac1.meth_out.value_counts().index[atac1.meth_out.value_counts() == 3]\n",
+    "atac1 = atac1.loc[atac1.meth_out.isin(meth_id_list)].reset_index(drop='index')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Filter each dataset to only contain methods that are in all 3 feature spaces\n",
+    "\n",
+    "atac2 = atac2.loc[~np.isnan(atac2['Overall Score']),:]\n",
+    "atac2['meth_out'] = ['_'.join(atac2[['Method', 'Output']].values[i]) for i in range(atac2.shape[0])]\n",
+    "\n",
+    "meth_id_list = atac2.meth_out.value_counts().index[atac2.meth_out.value_counts() == 3]\n",
+    "atac2 = atac2.loc[atac2.meth_out.isin(meth_id_list)].reset_index(drop='index')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "DESC_embed           3\n",
+       "BBKNN_graph          3\n",
+       "Conos_graph          3\n",
+       "SAUCIE_embed         3\n",
+       "SAUCIE_gene          3\n",
+       "ComBat_gene          3\n",
+       "Harmony_embed        3\n",
+       "LIGER_embed          3\n",
+       "Unintegrated_gene    3\n",
+       "Name: meth_out, dtype: int64"
+      ]
+     },
+     "execution_count": 6,
+     "metadata": {},
+     "output_type": "execute_result"
+    },
+    {
+     "data": {
+      "text/plain": [
+       "scVI_embed             3\n",
+       "Seurat v3 RPCA_gene    3\n",
+       "fastMNN_embed          3\n",
+       "ComBat_gene            3\n",
+       "MNN_gene               3\n",
+       "Seurat v3 CCA_gene     3\n",
+       "BBKNN_graph            3\n",
+       "scANVI*_embed          3\n",
+       "LIGER_embed            3\n",
+       "DESC_embed             3\n",
+       "scGen*_gene            3\n",
+       "Scanorama_embed        3\n",
+       "Harmony_embed          3\n",
+       "Scanorama_gene         3\n",
+       "fastMNN_gene           3\n",
+       "trVAE_embed            3\n",
+       "SAUCIE_gene            3\n",
+       "Conos_graph            3\n",
+       "SAUCIE_embed           3\n",
+       "Unintegrated_gene      3\n",
+       "Name: meth_out, dtype: int64"
+      ]
+     },
+     "execution_count": 6,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "atac1.meth_out.value_counts()\n",
+    "atac2.meth_out.value_counts()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "atac_filt = pd.concat([atac1,atac2])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "Feature Space\n",
+       "genes      0.406883\n",
+       "peaks      0.644815\n",
+       "windows    0.612147\n",
+       "Name: Bio conservation, dtype: float64"
+      ]
+     },
+     "execution_count": 8,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "atac_filt.groupby('Feature Space')['Bio conservation'].mean()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Remove unintegrated from this\n",
+    "atac_filt2 = atac_filt.loc[~atac_filt.meth_out.isin(['Unintegrated_gene'])].reset_index(drop='index')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "metadata": {
+    "scrolled": true
+   },
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "Feature Space\n",
+       "genes      0.395273\n",
+       "peaks      0.628008\n",
+       "windows    0.595843\n",
+       "Name: Bio conservation, dtype: float64"
+      ]
+     },
+     "execution_count": 10,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "atac_filt2.groupby('Feature Space')['Bio conservation'].mean()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "Feature Space\n",
+       "genes      0.655828\n",
+       "peaks      0.499953\n",
+       "windows    0.477366\n",
+       "Name: Batch Correction, dtype: float64"
+      ]
+     },
+     "execution_count": 11,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "atac_filt2.groupby('Feature Space')['Batch Correction'].mean()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Gene activity feature space has far lower bio conservation score than peaks and windows"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "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": "3.7.6"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}