tutorials/example_DMS.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from spectrae import Spectra, SpectraDataset\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from tqdm import tqdm"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# SPECTRA Deep Mutational Scan Tutorial\n",
    "\n",
    "### We take an example dataset from ProteinGym, A4_HUMAN_Seuma_2022.csv \n",
    "### We define a spectral property where two samples are similar if they share a mutation or not. \n",
    "### We define cross-split overlap as the proportion of mutations that are shared between samples in the train and test."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "class ProteinGym_Dataset(SpectraDataset):\n",
    "    \n",
    "    def parse(self, input_file):\n",
    "        return pd.read_csv(input_file)['mutant'].unique()\n",
    "    \n",
    "    def __len__(self):\n",
    "        return len(self.samples)\n",
    "    \n",
    "    def sample_to_index(self, sample):\n",
    "        if not hasattr(self, 'index_to_sequence'):\n",
    "            print(\"Generating index to sequence\")\n",
    "            self.index_to_sequence = {}\n",
    "            for i in tqdm(range(len(self))):\n",
    "                x = self.__getitem__(i)\n",
    "                self.index_to_sequence[x] = i\n",
    "\n",
    "        return self.index_to_sequence[sample] \n",
    "\n",
    "    def __getitem__(self, idx):\n",
    "        return self.samples[idx]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "class SPECTRA_DMS(Spectra):\n",
    "\n",
    "    def spectra_properties(self, sample_one, sample_two):\n",
    "        if sample_one != sample_two:\n",
    "            for mut in sample_one.split(':'):\n",
    "                if mut in sample_two:\n",
    "                    return 1\n",
    "            return 0\n",
    "        else:\n",
    "            return 1\n",
    "\n",
    "    def cross_split_overlap(self, train, test):\n",
    "\n",
    "        def flatten(l):\n",
    "            return [item for sublist in l for item in sublist]\n",
    "\n",
    "        one = flatten([i.split(':') for i in train])\n",
    "        two = flatten([i.split(':') for i in test])\n",
    "\n",
    "        def check_list(list, to_check):\n",
    "            for i in list:\n",
    "                if self.spectra_properties(i, to_check):\n",
    "                    return True\n",
    "            return False\n",
    "        \n",
    "        combined_mutations = list(set(one + two))\n",
    "\n",
    "        num_overlap = 0\n",
    "        for i in combined_mutations:\n",
    "            if check_list(one, i) and check_list(two, i):\n",
    "                num_overlap += 1\n",
    "        \n",
    "        return num_overlap/len(combined_mutations)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "PG_Dataset = ProteinGym_Dataset('A4_HUMAN_Seuma_2022.csv', 'A4_HUMAN_Seuma_2022')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "pg_spectra = SPECTRA_DMS(PG_Dataset, binary = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "File PGA4Human_precalculated_spectra_properties already exists, set force_recalculate to True to recalculate\n"
     ]
    }
   ],
   "source": [
    "pg_spectra.pre_calculate_spectra_properties('PGA4Human')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 0/109675455 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 109675455/109675455 [01:52<00:00, 975220.55it/s] \n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Stats for SPECTRA property graph (SPG)\n",
      "Number of nodes: 14268\n",
      "Number of edges: 2139791\n",
      "Number of connected components: 1\n",
      "\n",
      "\n",
      "SPG is not fully connected\n",
      "Component 0 is not fully connected\n",
      "Generating SPECTRA split for spectral parameter 0.0 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 14268\n",
      "Train size: 11414\tTest size: 2854\n",
      "Cross split overlap: 0.9565217391304348\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.0 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 14268\n",
      "Train size: 11414\tTest size: 2854\n",
      "Cross split overlap: 0.9644268774703557\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.0 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 14268\n",
      "Train size: 11414\tTest size: 2854\n",
      "Cross split overlap: 0.9802371541501976\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.05 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 2770\n",
      "Train size: 2216\tTest size: 554\n",
      "Cross split overlap: 0.8893280632411067\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.05 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 2779\n",
      "Train size: 2223\tTest size: 556\n",
      "Cross split overlap: 0.8774703557312253\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.05 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 2789\n",
      "Train size: 2231\tTest size: 558\n",
      "Cross split overlap: 0.8972332015810277\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.1 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 1735\n",
      "Train size: 1388\tTest size: 347\n",
      "Cross split overlap: 0.8458498023715415\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.1 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 1723\n",
      "Train size: 1378\tTest size: 345\n",
      "Cross split overlap: 0.8656126482213439\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.1 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 1721\n",
      "Train size: 1376\tTest size: 345\n",
      "Cross split overlap: 0.841897233201581\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.15 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 1302\n",
      "Train size: 1041\tTest size: 261\n",
      "Cross split overlap: 0.8221343873517787\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.15 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 1281\n",
      "Train size: 1024\tTest size: 257\n",
      "Cross split overlap: 0.841897233201581\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.15 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 1272\n",
      "Train size: 1017\tTest size: 255\n",
      "Cross split overlap: 0.782608695652174\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.2 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 1026\n",
      "Train size: 820\tTest size: 206\n",
      "Cross split overlap: 0.7154150197628458\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.2 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 1027\n",
      "Train size: 821\tTest size: 206\n",
      "Cross split overlap: 0.782608695652174\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.2 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 1013\n",
      "Train size: 810\tTest size: 203\n",
      "Cross split overlap: 0.7430830039525692\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.25 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 900\n",
      "Train size: 720\tTest size: 180\n",
      "Cross split overlap: 0.7035573122529645\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.25 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 879\n",
      "Train size: 703\tTest size: 176\n",
      "Cross split overlap: 0.6877470355731226\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.25 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 854\n",
      "Train size: 683\tTest size: 171\n",
      "Cross split overlap: 0.7391304347826086\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.3 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 744\n",
      "Train size: 595\tTest size: 149\n",
      "Cross split overlap: 0.6521739130434783\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.3 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 758\n",
      "Train size: 606\tTest size: 152\n",
      "Cross split overlap: 0.6363636363636364\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.3 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 762\n",
      "Train size: 609\tTest size: 153\n",
      "Cross split overlap: 0.6640316205533597\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.35 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 671\n",
      "Train size: 536\tTest size: 135\n",
      "Cross split overlap: 0.5968379446640316\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.35 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 668\n",
      "Train size: 534\tTest size: 134\n",
      "Cross split overlap: 0.6126482213438735\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.35 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 655\n",
      "Train size: 524\tTest size: 131\n",
      "Cross split overlap: 0.6007905138339921\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.4 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 582\n",
      "Train size: 465\tTest size: 117\n",
      "Cross split overlap: 0.5296442687747036\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.4 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 596\n",
      "Train size: 476\tTest size: 120\n",
      "Cross split overlap: 0.5889328063241107\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.4 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 573\n",
      "Train size: 458\tTest size: 115\n",
      "Cross split overlap: 0.5533596837944664\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.45 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 528\n",
      "Train size: 422\tTest size: 106\n",
      "Cross split overlap: 0.5454545454545454\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.45 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 536\n",
      "Train size: 428\tTest size: 108\n",
      "Cross split overlap: 0.5454545454545454\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.45 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 530\n",
      "Train size: 424\tTest size: 106\n",
      "Cross split overlap: 0.5217391304347826\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.5 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 455\n",
      "Train size: 364\tTest size: 91\n",
      "Cross split overlap: 0.4782608695652174\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.5 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 475\n",
      "Train size: 380\tTest size: 95\n",
      "Cross split overlap: 0.45454545454545453\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.5 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 475\n",
      "Train size: 380\tTest size: 95\n",
      "Cross split overlap: 0.4901185770750988\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.55 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 437\n",
      "Train size: 349\tTest size: 88\n",
      "Cross split overlap: 0.45849802371541504\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.55 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 435\n",
      "Train size: 348\tTest size: 87\n",
      "Cross split overlap: 0.45849802371541504\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.55 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 430\n",
      "Train size: 344\tTest size: 86\n",
      "Cross split overlap: 0.45849802371541504\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.6 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 397\n",
      "Train size: 317\tTest size: 80\n",
      "Cross split overlap: 0.4031620553359684\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.6 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 395\n",
      "Train size: 316\tTest size: 79\n",
      "Cross split overlap: 0.4268774703557312\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.6 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 392\n",
      "Train size: 313\tTest size: 79\n",
      "Cross split overlap: 0.4268774703557312\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.65 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 359\n",
      "Train size: 287\tTest size: 72\n",
      "Cross split overlap: 0.3557312252964427\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.65 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 354\n",
      "Train size: 283\tTest size: 71\n",
      "Cross split overlap: 0.3715415019762846\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.65 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 359\n",
      "Train size: 287\tTest size: 72\n",
      "Cross split overlap: 0.40711462450592883\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.7 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 325\n",
      "Train size: 260\tTest size: 65\n",
      "Cross split overlap: 0.3359683794466403\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.7 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 332\n",
      "Train size: 265\tTest size: 67\n",
      "Cross split overlap: 0.31620553359683795\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.7 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 344\n",
      "Train size: 275\tTest size: 69\n",
      "Cross split overlap: 0.3557312252964427\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.75 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 307\n",
      "Train size: 245\tTest size: 62\n",
      "Cross split overlap: 0.308300395256917\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.75 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 305\n",
      "Train size: 244\tTest size: 61\n",
      "Cross split overlap: 0.2885375494071146\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.75 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 294\n",
      "Train size: 235\tTest size: 59\n",
      "Cross split overlap: 0.3241106719367589\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.8 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 273\n",
      "Train size: 218\tTest size: 55\n",
      "Cross split overlap: 0.28063241106719367\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.8 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 280\n",
      "Train size: 224\tTest size: 56\n",
      "Cross split overlap: 0.25691699604743085\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.8 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 266\n",
      "Train size: 212\tTest size: 54\n",
      "Cross split overlap: 0.26877470355731226\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.85 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 246\n",
      "Train size: 196\tTest size: 50\n",
      "Cross split overlap: 0.2766798418972332\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.85 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 252\n",
      "Train size: 201\tTest size: 51\n",
      "Cross split overlap: 0.2450592885375494\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.85 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 246\n",
      "Train size: 196\tTest size: 50\n",
      "Cross split overlap: 0.22134387351778656\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.9 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 222\n",
      "Train size: 177\tTest size: 45\n",
      "Cross split overlap: 0.13438735177865613\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.9 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 225\n",
      "Train size: 180\tTest size: 45\n",
      "Cross split overlap: 0.18972332015810275\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.9 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 222\n",
      "Train size: 177\tTest size: 45\n",
      "Cross split overlap: 0.1857707509881423\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.95 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 187\n",
      "Train size: 149\tTest size: 38\n",
      "Cross split overlap: 0.09881422924901186\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.95 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 189\n",
      "Train size: 151\tTest size: 38\n",
      "Cross split overlap: 0.10276679841897234\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.95 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 195\n",
      "Train size: 156\tTest size: 39\n",
      "Cross split overlap: 0.11067193675889328\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 1.0 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 160\n",
      "Train size: 128\tTest size: 32\n",
      "Cross split overlap: 0.0\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 1.0 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 160\n",
      "Train size: 128\tTest size: 32\n",
      "Cross split overlap: 0.0\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 1.0 and dataset A4_HUMAN_Seuma_2022\n",
      "Deleted 0 nodes from the graph since they were connected to all other nodes\n",
      "Number of samples in independent set: 160\n",
      "Train size: 128\tTest size: 32\n",
      "Cross split overlap: 0.0\n",
      "\n",
      "\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "spectra_parameters = {'number_repeats': 3, \n",
    "                      'random_seed': [42, 44, 46],\n",
    "                      'spectral_parameters': [\"{:.2f}\".format(i) for i in np.arange(0, 1.05, 0.05)],\n",
    "                      'force_reconstruct': True,\n",
    "                                              }\n",
    "\n",
    "pg_spectra.generate_spectra_splits(**spectra_parameters)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x16239cd30>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "stats = pg_spectra.return_all_split_stats()\n",
    "plt.scatter(stats['SPECTRA_parameter'], stats['cross_split_overlap'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x16266abb0>"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(stats['SPECTRA_parameter'], stats['train_size'])\n",
    "plt.scatter(stats['SPECTRA_parameter'], stats['test_size']) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.9.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}