{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "f6c01484",
   "metadata": {},
   "source": [
    "Temporary notebook for manually scanning chromosomes for sequences of interest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "0608f91e",
   "metadata": {},
   "outputs": [],
   "source": [
    "seq_of_interest = \"GCAGATCTGCACATC\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "3c245151",
   "metadata": {},
   "outputs": [],
   "source": [
    "genome_dir = \"/home/a03-svincoff/DPACMAN/dpacman/data_files/raw/genomes/hg38\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "682098b6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['chr12', 'chr5', 'chr17', 'chr2', 'chr21', 'chr1', 'chrM', 'chr22', 'chr20', 'chr16', 'chr9', 'chr8', 'chr19', 'chr7', 'chr11', 'chr3', 'chr4', 'chr14', 'chr15', 'chr18', 'chrY', 'chr6', 'chrX', 'chr13', 'chr10'])\n"
     ]
    }
   ],
   "source": [
    "import json\n",
    "import os\n",
    "chrom_cache = {}\n",
    "for chrom_file in os.listdir(genome_dir):\n",
    "    chrom = chrom_file.split(\"hg38_\")[1].split(\".json\")[0]\n",
    "    with open(f\"{genome_dir}/{chrom_file}\", \"r\") as f:\n",
    "        chrom_cache[chrom] = json.load(f)\n",
    "\n",
    "print(chrom_cache.keys())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fd6cca79",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Testing sequence A: TAGCAGGATGTGT\n",
      "Testing sequence B: GCAGATCTGCACATC\n",
      "Testing sequence C: CGACACCTGACGCG\n",
      "Testing sequence D: CGCTATCCAGAGCG\n",
      "Testing sequence E: CGCGATGCTTCTCG\n",
      "Testing sequence F: CGGCTGGATTACCG\n",
      "Testing sequence G: CGAGAACATAGTCG\n",
      "Testing sequence H: CGGGGAAACGCCCG\n",
      "Testing sequence I: CGCCCAAAGCCGCG\n",
      "Testing sequence J: CGGAGGTAATGACG\n",
      "Testing sequence K: CGCACCGACTCACG\n",
      "Testing sequence L: CGGCCCTTTGCGCG\n",
      "Testing sequence M: CGCCGTTAGTGTCG\n"
     ]
    }
   ],
   "source": [
    "baker_sequences = {\n",
    "    \"A\": \"TAGCAGGATGTGT\",\n",
    "    \"B\": \"GCAGATCTGCACATC\",\n",
    "    \"C\": \"CGACACCTGACGCG\",\n",
    "    \"D\": \"CGCTATCCAGAGCG\",\n",
    "    \"E\": \"CGCGATGCTTCTCG\",\n",
    "    \"F\": \"CGGCTGGATTACCG\",\n",
    "    \"G\": \"CGAGAACATAGTCG\",\n",
    "    \"H\": \"CGGGGAAACGCCCG\",\n",
    "    \"I\": \"CGCCCAAAGCCGCG\",\n",
    "    \"J\": \"CGGAGGTAATGACG\",\n",
    "    \"K\": \"CGCACCGACTCACG\",\n",
    "    \"L\": \"CGGCCCTTTGCGCG\",\n",
    "    \"M\": \"CGCCGTTAGTGTCG\"\n",
    "}\n",
    "sorted_chroms = list(chrom_cache.keys())\n",
    "sorted_chroms = sorted(sorted_chroms, key = lambda x: int(x.split(\"chr\")[1]) if x.split(\"chr\")[1] not in [\"M\",\"X\",\"Y\"] else 0)\n",
    "\n",
    "for seq_letter, seq in baker_sequences.items():\n",
    "    print(f\"Testing sequence {seq_letter}: {seq}\")\n",
    "    for chrom in sorted_chroms:\n",
    "        chrom_dna = chrom_cache[chrom][\"dna\"].upper()\n",
    "        match_chroms=[]\n",
    "        try:\n",
    "            print(f\"\\tChrom {chrom} index of sequence {seq_letter} ({seq}): {chrom.index(seq)}\")\n",
    "            match_chroms+=[chrom]\n",
    "        except:\n",
    "            match_chroms = match_chroms\n",
    "    print(f\"\\tChrom {chrom} does not have sequence {seq_letter}({seq})\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "99078ac6",
   "metadata": {},
   "outputs": [],
   "source": [
    "#[m.start() for m in re.finditer(chr1_dna, 'GCAGATCTGCACATC')]"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "dnabind2",
   "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": "3.10.14"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}