ttv_project.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# TTV Project\n",
    "\n",
    "##### Identification of  *torque teno virus* genotypes in transplant donors and receivers <br>\n",
    "[=> OVERVIEW <=](https://docs.google.com/presentation/d/1ddYZr6uFXvWhx2wVmzzJ6F3a97x3xI6KONeIiZdLdm0/edit?usp=sharing)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1) Introduction <a name=\"introduction_1\"></a>\n",
    "Over the last few years, high throughput sequencing allowed the discovery and identification of new human-associated viruses. One of them is the *torque teno virus* (TTV). TTV is the dominant species inside the Annelovirus family, which itself is the dominant family of human-associated viruses. TTV has yet to be directly linked to a human disease, but we have many clues linking its abundance to various phenomena (for example in immunocompromised patients). TTV has a circular single-stranded DNA genome of approximately 3.8 kb. It's currently classified in 5 genogroup, containing 29 genotypes.\n",
    "\n",
    "Transplant receivers have a so-called \"optimal immunosupression state\", which is a state were the viral load is such that it minimize both reject risk and opportunistic infection risk *(Figure 1)*. One of the main hypothesis of this project is that the TTV viral load constitute a good indicator of the global viral load in transplant receivers, thus a good indicator of the immunosuppression state & reject risk."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Figure 1 : TTV load & immunosupression (adapted from De Vlaminck, Cell, 2013)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": []
    },
    {
     "data": {
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "### CODE ###\n",
    "display < ./Picture/optimal_immunosuppression.png;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2) Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Data consist of 20 pairs of transplant donors-receivers (for which TTV detection kit was positive). For each donor, 1 blood or plasma sample was drawn 1 day before donation. For the associated receiver, up to 3 blood or plasma samples were drawn after transplant at different time points (D35, D75 or D360). For each sample, DNA was extracted using a rolling circle amplification process (skewing RNA capture toward viral particles), then sequenced on Ion Proton platform (GenoLim - Limoges). A total of 66 samples were sequenced (20 donors and 46 receivers : 20 D35, 19 D75 and 7 D360) and submitted to the bioinformatics team for analysis. Correspondence between raw sequencing data and biological samples is given in this [spreadsheet](https://docs.google.com/spreadsheets/d/1KI94IUmesq1CZwdfSobA2qzEM8ZeqHWh8sR3QPar_K0/edit?usp=sharing) (correspondence was deduced from files \"*corresp_D-R_barecode_16062017.ppt*\" & \"*donneurs_receveurs.xlsx*\")."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.1) Raw reads"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### 106 FASTA files coming from 3 differents Ion Proton runs (GénoLim). <br>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "p1_r10.fastq.gz  p2_r36.fastq.gz   p31_r43.fastq.gz  p32_r40.fastq.gz\n",
      "p1_r11.fastq.gz  p2_r38.fastq.gz   p31_r44.fastq.gz  p32_r41.fastq.gz\n",
      "p1_r12.fastq.gz  p2_r39.fastq.gz   p31_r45.fastq.gz  p32_r42.fastq.gz\n",
      "p1_r13.fastq.gz  p2_r3.fastq.gz    p31_r46.fastq.gz  p32_r43.fastq.gz\n",
      "p1_r14.fastq.gz  p2_r40.fastq.gz   p31_r47.fastq.gz  p32_r44.fastq.gz\n",
      "p1_r16.fastq.gz  p2_r43.fastq.gz   p31_r48.fastq.gz  p32_r45.fastq.gz\n",
      "p1_r17.fastq.gz  p2_r46.fastq.gz   p31_r49.fastq.gz  p32_r46.fastq.gz\n",
      "p1_r19.fastq.gz  p2_r47.fastq.gz   p31_r50.fastq.gz  p32_r47.fastq.gz\n",
      "p1_r1.fastq.gz\t p2_r48.fastq.gz   p31_r51.fastq.gz  p32_r48.fastq.gz\n",
      "p1_r20.fastq.gz  p2_r49.fastq.gz   p31_r52.fastq.gz  p32_r49.fastq.gz\n",
      "p1_r22.fastq.gz  p2_r50.fastq.gz   p31_r56.fastq.gz  p32_r50.fastq.gz\n",
      "p1_r23.fastq.gz  p2_r51.fastq.gz   p31_r57.fastq.gz  p32_r51.fastq.gz\n",
      "p1_r24.fastq.gz  p2_r52.fastq.gz   p31_r58.fastq.gz  p32_r52.fastq.gz\n",
      "p1_r25.fastq.gz  p2_r54.fastq.gz   p31_r59.fastq.gz  p32_r56.fastq.gz\n",
      "p1_r2.fastq.gz\t p2_r55.fastq.gz   p31_r60.fastq.gz  p32_r57.fastq.gz\n",
      "p1_r3.fastq.gz\t p2_r56.fastq.gz   p31_r61.fastq.gz  p32_r58.fastq.gz\n",
      "p1_r5.fastq.gz\t p2_r63.fastq.gz   p31_r62.fastq.gz  p32_r59.fastq.gz\n",
      "p1_r6.fastq.gz\t p2_r64.fastq.gz   p31_r63.fastq.gz  p32_r60.fastq.gz\n",
      "p1_r7.fastq.gz\t p2_r96.fastq.gz   p31_r74.fastq.gz  p32_r61.fastq.gz\n",
      "p1_r8.fastq.gz\t p31_r33.fastq.gz  p31_r75.fastq.gz  p32_r62.fastq.gz\n",
      "p1_r9.fastq.gz\t p31_r36.fastq.gz  p31_r76.fastq.gz  p32_r63.fastq.gz\n",
      "p2_r21.fastq.gz  p31_r37.fastq.gz  p31_r77.fastq.gz  p32_r74.fastq.gz\n",
      "p2_r26.fastq.gz  p31_r38.fastq.gz  p32_r33.fastq.gz  p32_r75.fastq.gz\n",
      "p2_r27.fastq.gz  p31_r39.fastq.gz  p32_r36.fastq.gz  p32_r76.fastq.gz\n",
      "p2_r28.fastq.gz  p31_r40.fastq.gz  p32_r37.fastq.gz  p32_r77.fastq.gz\n",
      "p2_r29.fastq.gz  p31_r41.fastq.gz  p32_r38.fastq.gz\n",
      "p2_r35.fastq.gz  p31_r42.fastq.gz  p32_r39.fastq.gz\n"
     ]
    }
   ],
   "source": [
    "### CODE ###\n",
    "cd ~/Desktop/TTV/BISCEm/Data/Reads_raw\n",
    "ls *.fastq.gz;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Number of reads in raw files :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "| Sample           | Read_count |\n",
      "| ---------------- | ---------- |\n",
      "| p1_r2.fastq.gz   |  6 071 711 |\n",
      "| p1_r25.fastq.gz  |  5 239 434 |\n",
      "| p2_r48.fastq.gz  |  5 179 992 |\n",
      "| p2_r43.fastq.gz  |  4 704 575 |\n",
      "| p1_r1.fastq.gz   |  4 424 525 |\n",
      "| p2_r40.fastq.gz  |  4 370 049 |\n",
      "| p2_r46.fastq.gz  |  4 341 843 |\n",
      "| p2_r26.fastq.gz  |  4 306 879 |\n",
      "| p1_r13.fastq.gz  |  4 254 211 |\n",
      "| p2_r52.fastq.gz  |  4 139 308 |\n",
      "| p2_r96.fastq.gz  |  4 025 330 |\n",
      "| p2_r35.fastq.gz  |  3 541 846 |\n",
      "| p2_r51.fastq.gz  |  3 510 950 |\n",
      "| p1_r3.fastq.gz   |  3 247 795 |\n",
      "| p2_r21.fastq.gz  |  3 204 964 |\n",
      "| p2_r39.fastq.gz  |  3 077 039 |\n",
      "| p2_r50.fastq.gz  |  3 073 205 |\n",
      "| p2_r49.fastq.gz  |  3 059 484 |\n",
      "| p2_r27.fastq.gz  |  2 951 341 |\n",
      "| p2_r28.fastq.gz  |  2 748 693 |\n",
      "| p1_r11.fastq.gz  |  2 636 709 |\n",
      "| p2_r38.fastq.gz  |  2 612 764 |\n",
      "| p2_r36.fastq.gz  |  2 367 433 |\n",
      "| p1_r9.fastq.gz   |  2 246 469 |\n",
      "| p1_r24.fastq.gz  |  1 376 832 |\n",
      "| p2_r29.fastq.gz  |  1 350 724 |\n",
      "| p2_r64.fastq.gz  |  1 282 344 |\n",
      "| p1_r19.fastq.gz  |  1 258 156 |\n",
      "| p2_r47.fastq.gz  |  1 113 454 |\n",
      "| p1_r14.fastq.gz  |  1 062 606 |\n",
      "| p2_r56.fastq.gz  |  1 035 634 |\n",
      "| p1_r22.fastq.gz  |  1 003 164 |\n",
      "| p2_r55.fastq.gz  |    982 499 |\n",
      "| p2_r63.fastq.gz  |    937 971 |\n",
      "| p1_r12.fastq.gz  |    876 489 |\n",
      "| p32_r52.fastq.gz |    873 120 |\n",
      "| p1_r17.fastq.gz  |    864 546 |\n",
      "| p31_r59.fastq.gz |    807 090 |\n",
      "| p1_r23.fastq.gz  |    795 818 |\n",
      "| p1_r6.fastq.gz   |    775 339 |\n",
      "| p1_r16.fastq.gz  |    768 661 |\n",
      "| p32_r59.fastq.gz |    730 145 |\n",
      "| p2_r54.fastq.gz  |    701 780 |\n",
      "| p32_r51.fastq.gz |    696 660 |\n",
      "| p1_r7.fastq.gz   |    692 743 |\n",
      "| p32_r50.fastq.gz |    674 935 |\n",
      "| p32_r56.fastq.gz |    666 366 |\n",
      "| p32_r61.fastq.gz |    613 337 |\n",
      "| p1_r20.fastq.gz  |    588 923 |\n",
      "| p32_r48.fastq.gz |    586 715 |\n",
      "| p31_r40.fastq.gz |    582 261 |\n",
      "| p31_r56.fastq.gz |    544 412 |\n",
      "| p32_r62.fastq.gz |    543 329 |\n",
      "| p31_r52.fastq.gz |    533 922 |\n",
      "| p32_r40.fastq.gz |    508 750 |\n",
      "| p31_r62.fastq.gz |    467 554 |\n",
      "| p1_r5.fastq.gz   |    450 170 |\n",
      "| p31_r51.fastq.gz |    442 535 |\n",
      "| p31_r48.fastq.gz |    441 268 |\n",
      "| p32_r33.fastq.gz |    439 461 |\n",
      "| p32_r47.fastq.gz |    423 866 |\n",
      "| p31_r61.fastq.gz |    420 218 |\n",
      "| p32_r57.fastq.gz |    410 948 |\n",
      "| p32_r42.fastq.gz |    393 325 |\n",
      "| p32_r45.fastq.gz |    386 852 |\n",
      "| p31_r50.fastq.gz |    379 097 |\n",
      "| p31_r42.fastq.gz |    369 422 |\n",
      "| p32_r63.fastq.gz |    361 629 |\n",
      "| p31_r41.fastq.gz |    346 887 |\n",
      "| p31_r63.fastq.gz |    331 335 |\n",
      "| p31_r47.fastq.gz |    325 417 |\n",
      "| p31_r33.fastq.gz |    295 833 |\n",
      "| p32_r46.fastq.gz |    287 179 |\n",
      "| p31_r57.fastq.gz |    284 834 |\n",
      "| p32_r41.fastq.gz |    284 512 |\n",
      "| p31_r45.fastq.gz |    259 209 |\n",
      "| p32_r49.fastq.gz |    235 625 |\n",
      "| p31_r46.fastq.gz |    213 475 |\n",
      "| p32_r43.fastq.gz |    192 905 |\n",
      "| p31_r43.fastq.gz |    163 297 |\n",
      "| p31_r49.fastq.gz |    160 188 |\n",
      "| p32_r74.fastq.gz |    135 607 |\n",
      "| p31_r74.fastq.gz |     95 750 |\n",
      "| p32_r77.fastq.gz |     53 905 |\n",
      "| p32_r76.fastq.gz |     42 837 |\n",
      "| p31_r77.fastq.gz |     34 958 |\n",
      "| p32_r39.fastq.gz |     31 542 |\n",
      "| p32_r37.fastq.gz |     29 885 |\n",
      "| p32_r38.fastq.gz |     27 760 |\n",
      "| p31_r37.fastq.gz |     26 399 |\n",
      "| p32_r60.fastq.gz |     26 058 |\n",
      "| p31_r76.fastq.gz |     25 835 |\n",
      "| p31_r39.fastq.gz |     25 690 |\n",
      "| p31_r60.fastq.gz |     22 897 |\n",
      "| p31_r38.fastq.gz |     20 974 |\n",
      "| p31_r44.fastq.gz |     19 270 |\n",
      "| p31_r36.fastq.gz |     17 990 |\n",
      "| p32_r58.fastq.gz |     17 635 |\n",
      "| p32_r36.fastq.gz |     16 568 |\n",
      "| p31_r58.fastq.gz |     15 126 |\n",
      "| p32_r75.fastq.gz |     10 492 |\n",
      "| p31_r75.fastq.gz |      9 115 |\n",
      "| p32_r44.fastq.gz |      6 422 |\n",
      "| p1_r10.fastq.gz  |      1 426 |\n",
      "| p1_r8.fastq.gz   |        528 |\n",
      "| p2_r3.fastq.gz   |          7 |\n"
     ]
    }
   ],
   "source": [
    "### CODE ###\n",
    "cd ~/Desktop/TTV/BISCEm/Data/Reads_raw;\n",
    "for f in *fastq.gz;\n",
    "do lines=$(zgrep -Ec \"$\" $f);\n",
    "   printf $f;\n",
    "   printf '\\t'$(($lines/4))'\\n';\n",
    "done &>> temp.tsv;\n",
    "sed -i '1 i\\Sample\\tRead_count' temp.tsv;\n",
    "(head -1 temp.tsv && sort -nr -k2 <(tail -n +2 temp.tsv)) | csvlook;\n",
    "rm temp.tsv;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.2) Human genome"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The latest available human genome at the time of writing was retrieved from the genome reference consortium (GRC). "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### GRCh38.p11"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "f48d3166ede36c3381643dace60348b9  ./GCA_000001405.26_GRCh38.p11_genomic.fna.gz\n",
      "md5 check passed\n"
     ]
    }
   ],
   "source": [
    "### CODE ###\n",
    "# Download GRCh38.p11_genomic.fna.gz\n",
    "cd ~/Desktop/TTV/BISCEm/Data/Human;\n",
    "axel -q ftp://ftp.ncbi.nlm.nih.gov/genomes/all/GCA/000/001/405/\\\n",
    "GCA_000001405.26_GRCh38.p11/GCA_000001405.26_GRCh38.p11_genomic.fna.gz;\n",
    "# Download associated md5sum ftp file\n",
    "wget -q ftp://ftp.ncbi.nlm.nih.gov/genomes/all/GCA/000/001/405/\\\n",
    "GCA_000001405.26_GRCh38.p11/md5checksums.txt;\n",
    "# Perfom md5 check\n",
    "if grep \"$(md5sum GCA_000001405.26_GRCh38.p11_genomic.fna.gz | awk '{print $1'})\" md5checksums.txt;\n",
    "then echo \"md5 check passed\";\n",
    "     rm md5checksums.txt;\n",
    "else echo \"md5 check failed, file will be deleted\";\n",
    "     rm GCA_000001405.26_GRCh38.p11_genomic.fna.gz;\n",
    "     rm md5checksums.txt;\n",
    "fi;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.3) TTV genotypes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Retrieval sources :"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- NCBI taxonomy (29) : ID = 687331 <=> Alphatorquevirus : <br>\n",
    "Don't take those listed as \"simian\". <br>\n",
    "Manually retrieve them, then cat into > ~/Desktop/TTV/BISCEm/Data/TTV/ncbi_taxid_687331_ttv_20170622.fasta <br>\n",
    "\n",
    "\n",
    "- Reference genome (1) : <br>\n",
    "https://www.ncbi.nlm.nih.gov/nuccore/NC_015783.1 > ~/Desktop/TTV/BISCEm/Data/TTV/ncbi_ttv_ref_genome_20170622.fasta.gz <br>\n",
    "\n",
    "\n",
    "- Via entrez search terms (27) : <br>\n",
    "https://ncbiinsights.ncbi.nlm.nih.gov/2013/02/19/how-to-download-bacterial-genomes-using-the-entrez-api <br> https://www.ncbi.nlm.nih.gov/books/NBK179288"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "cd ~/Desktop/TTV/BISCEm/Data/TTV;\n",
    "\n",
    "# Retrieve TTV ID (!= taxo)\n",
    "wget -O ttv_genome.xml \"https://eutils.ncbi.nlm.nih.gov/entrez/\\\n",
    "eutils/esearch.fcgi?db=genome&term=torque+teno+virus\";\n",
    "id=$(grep -oP '<Id>\\K.*?(?=<\\/Id>)' ttv_genome.xml);\n",
    "\n",
    "# Use this ID to get all TTV genomes IDs from nuccore database\n",
    "wget -O ttv_genomes_ids.xml \"https://eutils.ncbi.nlm.nih.gov/entrez\\\n",
    "/eutils/elink.fcgi?dbfrom=genome&db=nuccore&id=$id\";\n",
    "grep -oP '<Id>\\K.*?(?=<\\/Id>)' ttv_genomes_ids.xml > ttv_genomes_ids.txt;\n",
    "sed -i '1d' ttv_genomes_ids.txt;\n",
    "\n",
    "# Filter out genomes IDs not associated with H.sapiens, then retrieve associated FASTA sequences\n",
    "while read line;\n",
    "do wget -O temp.txt \"https://eutils.ncbi.nlm.nih.gov/entrez/\\\n",
    "eutils/efetch.fcgi?db=nuccore&id=$line&retmode=text\"; \n",
    "   if grep --quiet \"sapiens\" temp.txt;\n",
    "   then echo $line >> ttv_genomes_ids_with_hsapiens.txt;\n",
    "   fi;\n",
    "done < ttv_genomes_ids.txt;\n",
    "while read line;\n",
    "do wget -O - \"https://eutils.ncbi.nlm.nih.gov/entrez/eutils/\\\n",
    "efetch.fcgi?db=nuccore&id=$line&rettype=fasta&retmode=text\" >> entrez_ttv_genomes_with_hsapiens_20170622.fasta;\n",
    "done < ttv_genomes_ids_with_hsapiens.txt;\n",
    "pigz entrez_ttv_genomes_with_hsapiens_20170622.fasta;\n",
    "\n",
    "# Clean all these temp files mess\n",
    "rm ttv_genome.xml ttv_genomes_ids.xml ttv_genomes_ids.txt temp.txt ttv_genomes_ids_with_hsapiens.txt;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then cat those 3 files in a single FASTA file (containing the 57 genotypes) :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "cat ncbi_taxid_687331_ttv_20170622.fasta.gz \\\n",
    "    ncbi_ttv_ref_genome_20170622.fasta.gz \\\n",
    "    entrez_ttv_genomes_with_hsapiens_20170622.fasta.gz > all_ttv_genotypes.fasta.gz;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "### 2.4) Microbial databases"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### ncbi_refseq_release vs ncbi_genome_refseq"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For viral & fungal DBs, we have the option to chose between ncbi_refseq_release \"RR\" and ncbi_genome_refseq \"GR\". <br>\n",
    "Below we try to understand which one to pick, using a viral DB exameple :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "# RR\n",
    "wget ftp://ftp.ncbi.nlm.nih.gov/refseq/release/viral/*.genomic.fna.gz;\n",
    "cat *.fna.gz > ncbi_refseq_release_viral.fna.gz;\n",
    "rm -f !(ncbi_refseq_release_viral.fna.gz);\n",
    "\n",
    "# GR\n",
    "curl 'ftp://ftp.ncbi.nlm.nih.gov/genomes/refseq/viral/assembly_summary.txt' \\\n",
    "| awk '{FS=\"\\t\"} !/^#/ {print $20}' \\\n",
    "| sed -r 's|(ftp://ftp.ncbi.nlm.nih.gov/genomes/all/.+/)(GCF_.+)|\\1\\2/\\2_genomic.fna.gz|' \\\n",
    "> list_viral.txt;\n",
    "sed -i 's/ftp:\\/\\//rsync:\\/\\//g' list_viral.txt;\n",
    "while read line;\n",
    "do rsync --quiet --times $line .;\n",
    "done < list_viral.txt;\n",
    "cat *.fna.gz > ncbi_genome_refseq_viral.fna.gz;\n",
    "rm -f !(ncbi_genome_refseq_viral.fna.gz);\n",
    "\n",
    "# Investigate difference between RR & GR\n",
    "zcat ncbi_refseq_release_viral.fna.gz | awk '/^>/ {print $1}' > ids_rr.txt;\n",
    "sed -i 's/>ref//g' ids_rr.txt;\n",
    "sed -i 's/|//g' ids_rr.txt;\n",
    "sort -u -o ids_rr.txt ids_rr.txt;\n",
    "zcat ncbi_genome_refseq_viral.fna.gz | awk '/^>/ {print $1}' > ids_gr.txt;\n",
    "sed -i 's/>//g' ids_gr.txt;\n",
    "sort -u -o ids_gr.txt ids_gr.txt;\n",
    "comm ids_rr.txt ids_gr.txt > comm_viral_RR_vs_GR.tsv;\n",
    "printf \"unique to rr :\\t\"; awk -F \"\\t\" '{if($1 != \"\") print $1}' comm_viral_RR_vs_GR.tsv | wc -l;\n",
    "printf \"unique to gr :\\t\"; awk -F \"\\t\" '{if($2 != \"\") print $2}' comm_viral_RR_vs_GR.tsv | wc -l;\n",
    "printf \"common to both :\\t\"; awk -F \"\\t\" '{if($3 != \"\") print $3}' comm_viral_RR_vs_GR.tsv | wc -l;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "218 are unique to RR <br>\n",
    "2 are unique to GR <br>\n",
    "9116 are common <br>\n",
    "The 2 seqs unique to GR are NC_025349.2 & NC_027348.1 <br>\n",
    "But they also exist in RR as NC_025349.3 & NC_027348.2 (more up-to-date versions) <br>\n",
    "Thus, use RR as it contains more stuff & up-to-date."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "#### 2.4.1) Viral"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "wget ftp://ftp.ncbi.nlm.nih.gov/refseq/release/viral/*.genomic.fna.gz;\n",
    "cat *.fna.gz > ncbi_refseq_release_viral.fna.gz;\n",
    "rm -f !(ncbi_refseq_release_viral.fna.gz);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.4.2) Fungal"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "wget ftp://ftp.ncbi.nlm.nih.gov/refseq/release/fungi/*.genomic.fna.gz;\n",
    "cat *.fna.gz > ncbi_refseq_release_fungi.fna.gz;\n",
    "rm -f !(ncbi_refseq_release_fungi.fna.gz);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.4.3) Bacterial"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Given that the NCBI refseq contain too much sequences (100+ Go) for the future blastn step to be done in a reasonable amount of computing time, we needed to reduce the size of the bacterial database. The choice was made to use The Human microbiome project (HMP) as a ressource. <br>\n",
    "In 2009, HMP compiled a database of microbial reference genomes (HMREFG), as described here : http://www.hmpdacc.org/HMREFG. This database contains archea, low-eucaryote, viral & bacterial, which are reported in the fasta header as follow : >ARCH, >EUKY, >VIRL, >BACT. The bacterial component of this database underwent a process of removing highly redundant, non HMP-sequenced reference genomes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "# Retrieve the HMREFG file, decompress & rename it\n",
    "cd ~/Desktop/TTV/BISCEm/Data/Bacterial;\n",
    "axel -q ftp://public-ftp.hmpdacc.org/HMREFG/all_seqs.fa.bz2;\n",
    "bzip2 -d all_seqs.fa.bz2;\n",
    "mv all_seqs.fa hmp_microorganisms.fasta;\n",
    "# Check which bacteria are included in this database\n",
    "grep '^>BAC' hmp_microorganisms.fasta \\\n",
    "| awk -F \"[[:space:]]\" '{print $2\"_\"$3}' \\\n",
    "| sort -u > hmp_bacterial_list.txt;\n",
    "head hmp_bacterial_list.txt;\n",
    "tail hmp_bacterial_list.txt;\n",
    "# Retrieve each bacteria header, then use seqtk to subsample all bacteria in the hmp_microorganisms.fasta\n",
    "grep '^>BAC' hmp_microorganisms.fasta \\\n",
    "| awk -F \"[[:space:]]\" '{print $1}' \\\n",
    "| sed 's/^>//g' > hmp_bacterial_header.list;\n",
    "seqtk subseq hmp_microorganisms.fasta hmp_bacterial_header.list > hmp_bacterial.fasta;\n",
    "# Compress data for storage and remove useless files\n",
    "pigz hmp_bacterial.fasta;\n",
    "pigz hmp_microorganisms.fasta;\n",
    "rm hmp_bacterial_header.list;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Ps : Bonus question : The HMREFG Reference Genome Database contains (quoting) \"1751 bacterial strains over 1253 species\". Let's try to verify this :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "zgrep '^>BAC' hmp_microorganisms.fasta.gz \\\n",
    "| awk -F '\\\\|' '{print $1}' \\\n",
    "| awk -F \"_\" '{print $2}' \\\n",
    "| sort -un \\\n",
    "| wc -l;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We are 17 strains short. This question was mailed to http://www.hmpdacc.org/outreach/feedback.php."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Ps bis : Get bacteria accession numbers :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "cd ~/Desktop/TTV/BISCEm/Data/Bacterial;\n",
    "awk -F \"[[:space:]]\" '{print $1}' hmp_bacterial_header.list | \\\n",
    "rev | sed 's/^|//' | cut -d'|' -f 1 | rev > hmp_bacterial_an.txt;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## 3) Reads filtering"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.1) Initial quality control"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### FastQC v0.11.5 :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#### CODE ###\n",
    "mkdir -p ~/Software/FastQC/Outputs/TTV ~/Desktop/TTV/BISCEm/Data/Reads_raw/QC;\n",
    "cd ~/Desktop/TTV/BISCEm/Data/Reads_raw;\n",
    "for f in *.fastq.gz;\n",
    "do fastqc --outdir ~/Software/FastQC/Outputs/TTV/ -f fastq $f;\n",
    "done >> ~/Desktop/TTV/BISCEm/Data/Reads_raw/QC/fastqc_verbose.txt;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### multiqc, version 0.9.dev0 : <br>\n",
    "*Ps* : Done for all combined runs & for each run separately, to highlight potential run biais."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "# For all runs\n",
    "cd ~/Software/MultiQC/Outputs;\n",
    "multiqc -i ttv -n ttv -v ~/Software/FastQC/Outputs/TTV/ >> multiqc_verbose.txt;\n",
    "cp -r ttv* ~/Desktop/TTV/BISCEm/Data/Reads_raw/QC;\n",
    "# Per run\n",
    "cd ~/Software/FastQC/Outputs/TTV;\n",
    "mkdir -p P1 P2 P31 P32;\n",
    "for f in p1_*; do mv $f P1; done;\n",
    "for f in p2_*; do mv $f P2; done;\n",
    "for f in p31_*; do mv $f P31; done;\n",
    "for f in p32_*; do mv $f P32; done;\n",
    "cd ~/Software/MultiQC/Outputs;\n",
    "multiqc -i ttv_p1 -n ttv_p1 -v ~/Software/FastQC/Outputs/TTV/P1;\n",
    "multiqc -i ttv_p2 -n ttv_p2 -v ~/Software/FastQC/Outputs/TTV/P2;\n",
    "multiqc -i ttv_p31 -n ttv_p31 -v ~/Software/FastQC/Outputs/TTV/P31;\n",
    "multiqc -i ttv_p32 -n ttv_p32 -v ~/Software/FastQC/Outputs/TTV/P32;\n",
    "cp -r *ttv_p* ~/Desktop/TTV/BISCEm/Data/Reads_raw/QC;\n",
    "cd ~/Software/FastQC/Outputs/TTV;\n",
    "mv P1/* P2/* P31/* P32/* .;\n",
    "rm -r P1 P2 P31 P32;\n",
    "# Visualize multiqc results\n",
    "cd ~/Desktop/TTV/BISCEm/Data/Reads_raw/QC;\n",
    "chromium-browser *.html;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Comments : <br>\n",
    "- The 5' per base sequence content seems skewed on the first 10 bases (i.e. not 25% for each nt), but this is expected for libraries produced by priming using random hexamers, which induce sequence composition biais on 5'. <br>\n",
    "- The over-represented sequence metric has no real sense for datasets with very few (100000 or less) reads. <br>\n",
    "- The per base sequence content is dirty toward the 3' end for our datasets, because there are a few (<1%%) sequences which are 200+ nts long, thus the ATGC% is obviously skewed."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.2) Filtering based on read quality"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Software : PRINSEQ-lite 0.20.4"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "biscem='/home/erwann/Desktop/TTV/BISCEm';\n",
    "cd $biscem/Data/Reads_raw;\n",
    "mkdir -p $biscem/Output/Reads_qual;\n",
    "out=$biscem'/Output/Reads_qual';\n",
    "echo $PRIN;\n",
    "for f in *.fastq.gz;\n",
    "do echo $f;\n",
    "   unpigz $f;\n",
    "   perl $PRIN -fastq ${f%.*} \\\n",
    "              -out_good $out'/'${f%%.*}'_qual' \\\n",
    "              -out_bad null \\\n",
    "              -trim_qual_type mean \\\n",
    "              -trim_qual_window 5 \\\n",
    "              -trim_qual_left 20 \\\n",
    "              -trim_qual_right 20 \\\n",
    "              -min_len 100 \\\n",
    "              -min_qual_mean 20 \\\n",
    "              -lc_method entropy \\\n",
    "              -lc_threshold 70 \\\n",
    "              -derep 1 \\\n",
    "              -derep_min 2;\n",
    "   pigz ${f%.*};\n",
    "   pigz  $out'/'${f%%.*}\"_qual\".fastq;\n",
    "done &>> $out'/prinseq.verbose';"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Order in which options are applied (see man prinseq) & what they do : <br>\n",
    "\n",
    "*Ps* : Triming options : A 5nts sliding windows checking mean score : -trim_qual_type mean & -trim_qual_window 5 <br>\n",
    "1) Trim 5' based on qual : -trim_qual_left 20 (Phred score) <br>\n",
    "2) Trim 3' based on qual : -trim_qual_right 20 <br>\n",
    "3) Discard reads shorter than 100 nts : -min_len 100 <br>\n",
    "4) Filter sequence with mean qual score below 20 : -min_qual_mean 20 <br>\n",
    "5) Discard low-complexity reads : -lc_method entropy & -lc_threshold 70 (ala Vlaminck) <br>\n",
    "6) Remove exact duplicates : -derep 1 & -derep_min 2 <br>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3.2.1) QC"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Goal : Check if there is a notable improvement compared to QC done on raw reads."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "fastqc='/home/erwann/Software/FastQC/Outputs';\n",
    "# FastQC\n",
    "mkdir -p $fastqc/TTV_qual \\\n",
    "         $out/QC;\n",
    "cd $out;\n",
    "for f in *.fastq.gz;\n",
    "do fastqc --outdir $fastqc/TTV_qual/ -f fastq $f;\n",
    "done &>> QC/fastqc_verbose.txt;\n",
    "# MultiQC\n",
    "cd $fastqc/TTV_qual;\n",
    "mkdir -p P1 P2 P31 P32;\n",
    "for f in p1_*; do mv $f P1; done;\n",
    "for f in p2_*; do mv $f P2; done;\n",
    "for f in p31_*; do mv $f P31; done;\n",
    "for f in p32_*; do mv $f P32; done;\n",
    "cd ~/Software/MultiQC/Outputs;\n",
    "multiqc -i TTV_qual_p1 -n TTV_qual_p1 -v $fastqc/TTV_qual/P1;\n",
    "multiqc -i TTV_qual_p2 -n TTV_qual_p2 -v $fastqc/TTV_qual/P2;\n",
    "multiqc -i TTV_qual_p31 -n TTV_qual_p31 -v $fastqc/TTV_qual/P31;\n",
    "multiqc -i TTV_qual_p32 -n TTV_qual_p32 -v $fastqc/TTV_qual/P32;\n",
    "cp -r TTV_qual_* $out/QC;\n",
    "cd $fastqc/TTV_qual;\n",
    "mv P1/* P2/* P31/* P32/* .;\n",
    "rm -r P1 P2 P31 P32;\n",
    "# Visualize\n",
    "cd $out/QC;\n",
    "chromium-browser *.html;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3.2.2) Merge FASTQ files belonging to the same sample"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "First, create a backup folder with unmerged FASTQ, then use bash script to rename :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "# Backup just in case\n",
    "cd $out;\n",
    "mkdir -p Before_sample_merging;\n",
    "cp *.fastq.gz Before_sample_merging;\n",
    "# Bash rename\n",
    "bash rename_to_sample.sh;\n",
    "rm p*.*fastq.gz;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "*Ps* : Content of rename_to_sample.sh :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "cat p31_r74_qual.fastq.gz p32_r74_qual.fastq.gz > d1_qual.fastq.gz;\n",
    "cat p31_r75_qual.fastq.gz p32_r75_qual.fastq.gz > d3_qual.fastq.gz;\n",
    "mv p1_r2_qual.fastq.gz d4_qual.fastq.gz;\n",
    "...\n",
    "mv p2_r54_qual.fastq.gz r44_qual.fastq.gz;\n",
    "mv p2_r55_qual.fastq.gz r45_qual.fastq.gz;\n",
    "cat p2_r56_qual.fastq.gz p31_r33_qual.fastq.gz p32_r33_qual.fastq.gz \\\n",
    "    p31_r63_qual.fastq.gz p32_r63_qual.fastq.gz > r46_qual.fastq.gz;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Number of reads in quality filtered files :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "| Sample            | Read_count |\n",
      "| ----------------- | ---------- |\n",
      "| r36_qual.fastq.gz |  4 165 005 |\n",
      "| d4_qual.fastq.gz  |  3 892 373 |\n",
      "| r38_qual.fastq.gz |  3 626 523 |\n",
      "| r25_qual.fastq.gz |  3 257 880 |\n",
      "| r33_qual.fastq.gz |  3 237 019 |\n",
      "| d17_qual.fastq.gz |  3 217 460 |\n",
      "| r43_qual.fastq.gz |  2 978 223 |\n",
      "| r42_qual.fastq.gz |  2 905 854 |\n",
      "| r30_qual.fastq.gz |  2 802 911 |\n",
      "| r5_qual.fastq.gz  |  2 775 013 |\n",
      "| r1_qual.fastq.gz  |  2 490 595 |\n",
      "| r41_qual.fastq.gz |  2 358 465 |\n",
      "| r40_qual.fastq.gz |  2 288 172 |\n",
      "| r4_qual.fastq.gz  |  2 263 400 |\n",
      "| r39_qual.fastq.gz |  2 149 094 |\n",
      "| r2_qual.fastq.gz  |  2 096 159 |\n",
      "| r16_qual.fastq.gz |  1 977 498 |\n",
      "| r28_qual.fastq.gz |  1 782 717 |\n",
      "| d14_qual.fastq.gz |  1 764 340 |\n",
      "| r46_qual.fastq.gz |  1 753 767 |\n",
      "| r29_qual.fastq.gz |  1 669 163 |\n",
      "| r18_qual.fastq.gz |  1 438 737 |\n",
      "| r13_qual.fastq.gz |  1 291 809 |\n",
      "| r26_qual.fastq.gz |  1 252 085 |\n",
      "| r32_qual.fastq.gz |  1 217 526 |\n",
      "| r3_qual.fastq.gz  |  1 194 441 |\n",
      "| r9_qual.fastq.gz  |  1 122 529 |\n",
      "| d16_qual.fastq.gz |  1 044 648 |\n",
      "| r17_qual.fastq.gz |  1 005 644 |\n",
      "| r27_qual.fastq.gz |    912 815 |\n",
      "| r24_qual.fastq.gz |    886 380 |\n",
      "| r6_qual.fastq.gz  |    857 742 |\n",
      "| r23_qual.fastq.gz |    809 799 |\n",
      "| r35_qual.fastq.gz |    807 015 |\n",
      "| d20_qual.fastq.gz |    786 334 |\n",
      "| r19_qual.fastq.gz |    761 578 |\n",
      "| r21_qual.fastq.gz |    748 828 |\n",
      "| r11_qual.fastq.gz |    724 772 |\n",
      "| r37_qual.fastq.gz |    666 066 |\n",
      "| r20_qual.fastq.gz |    614 008 |\n",
      "| r14_qual.fastq.gz |    548 831 |\n",
      "| d19_qual.fastq.gz |    542 390 |\n",
      "| r44_qual.fastq.gz |    523 304 |\n",
      "| r8_qual.fastq.gz  |    481 818 |\n",
      "| d12_qual.fastq.gz |    429 351 |\n",
      "| r45_qual.fastq.gz |    428 217 |\n",
      "| r15_qual.fastq.gz |    405 803 |\n",
      "| r10_qual.fastq.gz |    405 410 |\n",
      "| r22_qual.fastq.gz |    296 997 |\n",
      "| r12_qual.fastq.gz |    231 167 |\n",
      "| d10_qual.fastq.gz |    210 023 |\n",
      "| d11_qual.fastq.gz |    157 168 |\n",
      "| d1_qual.fastq.gz  |    125 377 |\n",
      "| d8_qual.fastq.gz  |     45 694 |\n",
      "| d5_qual.fastq.gz  |     40 106 |\n",
      "| r34_qual.fastq.gz |     21 697 |\n",
      "| r7_qual.fastq.gz  |     18 819 |\n",
      "| r31_qual.fastq.gz |     12 308 |\n",
      "| d3_qual.fastq.gz  |      7 404 |\n",
      "| d15_qual.fastq.gz |        956 |\n",
      "| d13_qual.fastq.gz |        104 |\n"
     ]
    }
   ],
   "source": [
    "### CODE ###\n",
    "cd ~/Desktop/TTV/BISCEm/Output/Reads_qual;\n",
    "for f in *fastq.gz;\n",
    "do lines=$(zgrep -Ec \"$\" $f);\n",
    "   printf $f;\n",
    "   printf '\\t'$(($lines/4))'\\n';\n",
    "done &>> temp.tsv;\n",
    "sed -i '1 i\\Sample\\tRead_count' temp.tsv;\n",
    "(head -1 temp.tsv && sort -nr -k2 <(tail -n +2 temp.tsv)) | csvlook;\n",
    "rm temp.tsv;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [],
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "display < ./Picture/read_count_qual.png;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "*Ps* : We should have 66 samples (20 D, 46 R), but 5 of them (D2, D6, D7, D9 & D18) have no sequenced reads. <br>\n",
    "Thus at this stage, we have 61.fastq.gz files for the remaining 61 samples."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.3) Human reads removal"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3.3.1) Load data on genotoul :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "cd ~/Desktop/TTV/BISCEm/Output;\n",
    "sudo scp -r Reads_qual escaon@genotoul.toulouse.inra.fr:/home/escaon/save/TTV;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Using BWA-0.7.15 to map reads on the human genome."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3.3.2) Build BWA index :"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Content of bwa_idx.sh :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "#!/bin/bash\n",
    "#$ -q workq\n",
    "#$ -M erwan.scaon@unilim.fr\n",
    "#$ -m bea\n",
    "#$ -l mem=48G\n",
    "#$ -l h_vmem=60G\n",
    "#$ -N GRCh38_bwa_idx\n",
    "#$ -o /home/escaon/work/TTV/bwa_idx_o.txt\n",
    "#$ -e /home/escaon/work/TTV/bwa_idx_e.txt\n",
    "\n",
    "BWA='/usr/local/bioinfo/src/bwa/bwa-0.7.15/bwa';\n",
    "path='/home/escaon/save/TTV/Data';\n",
    "\n",
    "$BWA index -p GRCh38_p11_bwtsw -a bwtsw $path/GCA_000001405.26_GRCh38.p11_genomic.fna;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3.3.3) Map reads :"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Content of bwa_mem.sh (keep only unmapped) :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "#!/bin/bash\n",
    "#$ -q workq\n",
    "#$ -M erwan.scaon@unilim.fr\n",
    "#$ -m bea\n",
    "#$ -l mem=8G\n",
    "#$ -l h_vmem=12G\n",
    "#$ -N TTV_bwa_mem\n",
    "#$ -o /home/escaon/work/TTV/ttv_bwa_mem_20170907_o.txt\n",
    "#$ -e /home/escaon/work/TTV/ttv_bwa_mem_20170907_e.txt\n",
    "\n",
    "# qsub -pe parallel_smp 8 bwa_mem.sh\n",
    "\n",
    "BWA='/usr/local/bioinfo/src/bwa/bwa-0.7.15/bwa';\n",
    "save='/home/escaon/save/TTV';\n",
    "work='/home/escaon/work/TTV';\n",
    "nt=8;\n",
    "\n",
    "mkdir -p $work/Reads_qual_hm;\n",
    "cd $save/Reads_qual;\n",
    "for f in *.fastq.gz;\n",
    "do $BWA mem -t $nt \\\n",
    "            -R '@RG\\tID:'${f%%.*}'\\tLB:'${f%%.*}'\\tPL:IONTORRENT\\tPU:GenoLim\\tSM:'${f%%.*} \\\n",
    "            $work/Bwa_idx/GRCh38_p11_bwtsw \\\n",
    "            <(unpigz -c $f) |\\\n",
    "            samtools view -bh -f 4 \\\n",
    "            > $work/Reads_qual_hm/${f%%.*}.bam;\n",
    "done;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3.3.4) Validate BAMs :"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Content of picard_validate.sh :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "#!/bin/bash\n",
    "#$ -q workq\n",
    "#$ -M erwan.scaon@unilim.fr\n",
    "#$ -m bea\n",
    "#$ -l mem=32G\n",
    "#$ -l h_vmem=40G\n",
    "#$ -N TTV_picard_validate\n",
    "#$ -o /home/escaon/work/TTV/ttv_picard_val_o.txt\n",
    "#$ -e /home/escaon/work/TTV/ttv_picard_val_e.txt\n",
    "\n",
    "work='/home/escaon/work/TTV';\n",
    "module load bioinfo/Java8;\n",
    "PICARD='/usr/local/bioinfo/src/picard-tools/picard-tools-2.1.1/picard.jar';\n",
    "\n",
    "cd $work/Reads_qual_hm;\n",
    "for f in *.bam;\n",
    "do echo $f; java -jar $PICARD ValidateSamFile I=$f MODE=SUMMARY;\n",
    "done;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3.3.5) Sort & index BAMs :"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Content of samtools_sort_idx.sh :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "#!/bin/bash\n",
    "#$ -q workq\n",
    "#$ -M erwan.scaon@unilim.fr\n",
    "#$ -m bea\n",
    "#$ -l mem=40G\n",
    "#$ -l h_vmem=48G\n",
    "#$ -N TTV_samtools_sort_idx\n",
    "#$ -o /home/escaon/work/TTV/ttv_samtools_sortidx_o.txt\n",
    "#$ -e /home/escaon/work/TTV/ttv_samtools_sortidx_e.txt\n",
    "\n",
    "work='/home/escaon/work/TTV';\n",
    "\n",
    "cd $work/Reads_qual_hm;\n",
    "for f in *.bam;\n",
    "do samtools sort -o ${f%.*}\"_sorted.bam\" $f;\n",
    "done;\n",
    "for f in *_sorted.bam;\n",
    "do samtools index $f;\n",
    "done;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3.3.6) BAM to FASTA :"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Content of samtools_fasta.sh :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "#!/bin/bash\n",
    "#$ -q workq\n",
    "#$ -M erwan.scaon@unilim.fr\n",
    "#$ -m bea\n",
    "#$ -l mem=32G\n",
    "#$ -l h_vmem=40G\n",
    "#$ -N TTV_samtools_fasta\n",
    "#$ -o /home/escaon/work/TTV/ttv_samtools_fasta_o.txt\n",
    "#$ -e /home/escaon/work/TTV/ttv_samtools_fasta_e.txt\n",
    "\n",
    "work='/home/escaon/work/TTV';\n",
    "\n",
    "cd $work/Reads_qual_hm;\n",
    "for f in *_sorted.bam;\n",
    "do samtools fasta $f > ${f%_*}.fasta;\n",
    "done;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3.3.7) DL these FASTA files locally :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "sudo scp escaon@genotoul.toulouse.inra.fr:/home/escaon/work/TTV/Reads_qual_hm/*.fasta \\\n",
    "         /home/erwann/Desktop/TTV/BISCEm/Output/Reads_qual_hm;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Number of reads in filtered files (*Ps* : Could have been checked in BAM (samtools view -c toto.bam)) : <br>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "cd ~/Desktop/TTV/BISCEm/Output/Reads_qual_hm;\n",
    "for f in *.fasta;\n",
    "do printf $f'\\t';\n",
    "   grep -c '^>' $f;\n",
    "done;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.4) 5' UTR reads removal"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Rappel de l'objectif ici : Sachant que les génotypes de TTV partagent une zone en 5' qui est fortement conservée, il serait bon de ne pas tenir compte des reads qui s'alignent pour une majeur partie de leur longueur sur cette zone 5', pour éviter de créer des biais au niveau du compte des reads assignés à chaque génotype (en gros on préfère se focaliser sur les reads vraiment discriminants).\n",
    "\n",
    "Ok, donc je pars sur ce consensus +++ conservé de 93nts. Cf : /home/erwann/Desktop/TTV/BISCEm/Data/TTV/Genomes_MSA/MultAlin result.html\n",
    "\n",
    "*Nb* : En 3' de ce consensus, on retrouve directement pas mal de variabilité entre les génotypes. C'est moins vrai en 5', je dirais que les 67 nts en amont du consensus sont relativement conservés, donc pas forcément très discriminants pour nos objectifs.\n",
    "\n",
    "Je pense donc appliquer la logique suivante :\n",
    "Pour chaque jeu de données après filtre sur la qualité et soustraction des reads humain, aligner les reads restants contre le consensus de 93 nts.\n",
    "Si un read possède un hit de 25 ou plus nts dans cette zone de 93nts alors :\n",
    "Si ce read possède, en 3' du consensus, 50 ou plus nts non-impliqués dans le hit , on le conserve (car ces 50+ nts seront discriminants).\n",
    "Si ce read possède, en 5' du consensus + 67 nts (pour tenir compte de la partie environ conservée), 50 ou plus nts non-impliqués dans le hit, on le conserve (car ces 50+ nts seront discriminants).\n",
    "Si le read n'a pas assez de nts discriminants en 3' ou en 5' du consensus, alors il sera retiré du jeu de données."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Content of discard_5prime_consensus_reads.py :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "#!/usr/bin/env python\n",
    "import os\n",
    "import sys\n",
    "\n",
    "usage = '\\t --------\\n' \\\n",
    "        '\\t| usage  : python discard_5prime_consensus_reads file_1\\n' \\\n",
    "        '\\t| input  : file_1 = blastn_vs_5prime_consensus.tsv\\n' \\\n",
    "        '\\t| output : a file containing read names to discard\\n' \\\n",
    "        '\\t --------'\n",
    "\n",
    "if len(sys.argv) != 2:\n",
    "    print(usage)\n",
    "    sys.exit()\n",
    "\n",
    "query_dic = {}\n",
    "with open(sys.argv[1], 'r') as tsv:\n",
    "    for row in tsv:\n",
    "        columns = row.split('\\t')\n",
    "        qseqid, length, qstart, qend, sstart, send, qlen, sstrand = columns[0], int(columns[3]), int(\n",
    "            columns[6]), int(columns[7]), int(columns[8]), int(columns[9]), int(columns[12]), columns[14].rstrip()\n",
    "        if not qseqid in query_dic:\n",
    "            if length >= 25:\n",
    "                if sstrand == 'plus':\n",
    "                    five_prime_overhead = (qstart - 1) - (sstart - 1) - 67\n",
    "                    three_prime_overhead = (qlen - qend) - (93 - send)\n",
    "                elif sstrand == 'minus':\n",
    "                    five_prime_overhead = (qlen - qend) - (send - 1) - 67\n",
    "                    three_prime_overhead = (qstart - 1) - (93 - sstart)\n",
    "                else:\n",
    "                    print('------------ERROR------------')\n",
    "                if five_prime_overhead >= 50 or three_prime_overhead >= 50:\n",
    "                    query_dic[qseqid] = 'keep'\n",
    "                else:\n",
    "                    query_dic[qseqid] = 'discard'\n",
    "\n",
    "file = sys.argv[1].split('/')\n",
    "sample = file[-1].split('.')[0]\n",
    "\n",
    "with open(sample + '.discard', 'a') as out:\n",
    "    for k, v in query_dic.items():\n",
    "        if v == 'discard':\n",
    "            out.write(k + \"\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Apply this filtering step :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "biscem='/home/erwann/Desktop/TTV/BISCEm';\n",
    "blast='/home/erwann/Software/Ncbi_blast_2.6.0+/bin';\n",
    "mkdir -p $biscem/Output/Reads_qual_hm_fivep;\n",
    "cd $biscem/Output/Reads_qual_hm;\n",
    "for f in *.fasta;\n",
    "do $blast/blastn -subject $biscem/Data/TTV/ttv_5prime_consensus.fasta \\\n",
    "                 -query $f \\\n",
    "                 -reward 1 \\\n",
    "                 -penalty -3 \\\n",
    "                 -word_size 12 \\\n",
    "                 -gapopen 5 \\\n",
    "                 -gapextend 2 \\\n",
    "                 -evalue 0.0001 \\\n",
    "                 -dust \"no\" \\\n",
    "                 -perc_identity 90 \\\n",
    "                 -culling_limit 2 \\\n",
    "                 -out ${f%.*}.tsv \\\n",
    "                 -outfmt \"6 qseqid sseqid pident length mismatch gapopen qstart \\\n",
    "                          qend sstart send evalue bitscore qlen nident sstrand\";\n",
    "                 python $biscem/Scripts/discard_5prime_consensus_reads.py ${f%.*}.tsv;\n",
    "                 grep '^>' $f | sort > ${f%.*}.headers; sed -i 's/^>//g' ${f%.*}.headers;\n",
    "                 sort ${f%.*}.discard -o ${f%.*}.discard;\n",
    "                 comm -23 ${f%.*}.headers ${f%.*}.discard > ${f%.*}.keepers;\n",
    "                 seqtk subseq $f ${f%.*}.keepers > $biscem/Output/Reads_qual_hm_fivep/$f;\n",
    "done;\n",
    "rm *.tsv *.headers *.discard *.keepers;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Number of reads in filtered files :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "cd ~/Desktop/TTV/BISCEm/Output/Reads_qual_hm_fivep;\n",
    "for f in *.fasta;\n",
    "do printf $f'\\t';\n",
    "   grep -c '^>' $f;\n",
    "done;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Upload them on genotoul & CALI :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "sudo scp -r /home/erwann/Desktop/TTV/BISCEm/Output/Reads_qual_hm_fivep \\\n",
    "            escaon@genotoul.toulouse.inra.fr:/home/escaon/save/TTV;\n",
    "sudo scp -r /home/erwann/Desktop/TTV/BISCEm/Output/Reads_qual_hm_fivep \\\n",
    "            scaonp01@cali.unilim.fr://home/scaonp01/TTV;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4) TTV taxonomical analysis"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It was decided to focus on Orf1 for this analysis, based on 65% homology (Maggi, etc...)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.1) Retrieve Orf1 sequences for our 57 genotypes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Manual inspection of the 57 genotypes entries for orf1 :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "# Open NCBI accession webpage for all TTV genotypes\n",
    "cd /home/erwann/Desktop/TTV/BISCEm/Data/TTV;\n",
    "while read line;\n",
    "do chromium-browser https://www.ncbi.nlm.nih.gov/nuccore/$line;\n",
    "done < an_all_ttv_genotypes.txt;\n",
    "# Create an empty BED file to store the infos that we are going to retrieve manually :-(\n",
    "cd /home/erwann/Desktop/TTV/BISCEm/Data/TTV;\n",
    "while read line;\n",
    "do touch ../../Orf1/$line.bed;\n",
    "done < an_all_ttv_genotypes.txt;\n",
    "# Then fill the 57 BED files manually based on webpage info (start pos -1 plz) e.g. AB038621.1.bed => AB038621.1 665 2867\n",
    "# The entries listed below have no annotated CDS for Orf1 :\n",
    "# AX025718.1, AX025830.1, AX174942.1, KP343839.1 and NC_015783.1\n",
    "# For them, go to https://www.ebi.ac.uk/Tools/psa/genewise/ & align this Orf1 protein sequence against them (https://www.ncbi.nlm.nih.gov/nuccore/NC_002076.2). They all have a hit <=> orf1, thus put this info in their BED\n",
    "# Then use seqtk to retrieve Orf1 sequences based on BEDs\n",
    "cd /home/erwann/Desktop/TTV/BISCEm/Data/TTV/All_ttv_genotypes_one_per_fasta;\n",
    "for f in *.fasta;\n",
    "do seqtk subseq $f ../Orf1/${f%.*}.bed > ../Orf1/${f%.*}.fasta;\n",
    "done;\n",
    "# Check that our resulting Orf1 seqs len are / by 3\n",
    "cd ../Orf1;\n",
    "for f in *.fasta;\n",
    "do printf $f\" : \"; grep -v \">\" $f \\\n",
    "   | wc \\\n",
    "   | awk '{print ($3-$1)/3}';\n",
    "done;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.2) Create taxo groups based on Orf1 homology"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Voici les résultats de la phylogénie orf1 (alignement mafft+rooted UPGMA tree). <br>\n",
    "En comptant 65% de similitude on part sur 30 groupes : *cf* groupes-pim.html (percentage identity matrix) et figure ci-dessous :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "display < ./Picture/taxo_tree.png;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5 Assign reads to microbial DBs & TTV genotypes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### blast 2.6.0+"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Fungal & bacterial blastn output needs (definie outfmt 6 fields) : <br>\n",
    "qlen # Query sequence length (needed to filter hits shorter than 90% of qlen) <br>\n",
    "saccver # Subject accession.version (in case of future need for eutils => access to taxo infos) <br>\n",
    "Which translate into : -outfmt \"6 qseqid sseqid pident length mismatch gapopen qstart qend sstart send evalue bitscore qlen\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Viral blastn output needs : <br>\n",
    "We need additional outfmt 6 fields to parse for Orf1 : <br>\n",
    "qstart # Start of alignment in query <br>\n",
    "qend # End of alignment in query <br>\n",
    "slen # Subject sequence length <br>\n",
    "sstart # Start of alignment in subject <br>\n",
    "send # End of alignment in subject <br>\n",
    "sstrand # Subject Strand <br>\n",
    "Which translate into : -outfmt \"6 qseqid sseqid pident length mismatch gapopen qstart qend sstart send evalue bitscore qlen qstart qend slen sstart send sstrand\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.1) Create databases"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 5.1.1) Viral DBs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "# Check intersection between viral_rr & our_57_ttv_genotypes\n",
    "data='/home/erwann/Desktop/TTV/BISCEm/Data';\n",
    "cd $data/Viral;\n",
    "sort -u -o ids_ncbi_refseq_release_viral.txt ids_ncbi_refseq_release_viral.txt;\n",
    "cd $data/TTV;\n",
    "sort -u -o an_all_ttv_genotypes.txt an_all_ttv_genotypes.txt;\n",
    "comm an_all_ttv_genotypes.txt $data/Viral/ids_ncbi_refseq_release_viral.txt > ttvG_vs_viralRefSeq_intersection.tsv;\n",
    "\n",
    "cd $data/TTV;\n",
    "zgrep -i 'complete genome' all_ttv_genotypes.fasta.gz > complete_genomes.txt;\n",
    "sed -i 's/>//g' complete_genomes.txt;\n",
    "awk -F ' ' '{print $1}' complete_genomes.txt > an_complete_genomes.txt;\n",
    "rm complete_genomes.txt;\n",
    "\n",
    "# All \"NC\" genomes (18) amongst our 57 ttv genotypes are complete genomes and are present in viral_rr. We have 4 additional genomes labeled as complete amongst our 57, which are : KP343839.1, KP036971.1, AY823989.1 & AY823988.1.\n",
    "# Let's add them to viral_rr to create future rr_viral_plus_complete_genome (create a file call add_those_seq_to_refseq.txt containing those 4 an).\n",
    "\n",
    "seqtk subseq all_ttv_genotypes.fasta.gz add_those_seq_to_refseq.txt> add_those_complete_genome_to_rr.fasta;\n",
    "pigz add_those_complete_genome_to_rr.fasta;\n",
    "cat add_those_complete_genome_to_rr.fasta.gz $data/Viral/ncbi_refseq_release_viral.fna.gz > ncbi_refseq_release_viral_plus_ttv_complete_genomes.fasta.gz;\n",
    "\n",
    "blast='/home/erwann/Software/Ncbi_blast_2.6.0+/bin/';\n",
    "unpigz ncbi_refseq_release_viral_plus_ttv_complete_genomes.fasta.gz;\n",
    "$blast/makeblastdb -in ncbi_refseq_release_viral_plus_ttv_complete_genomes.fasta -input_type fasta -dbtype nucl -out $data/Blast_DB/rr_viral_plus_complete_genome;\n",
    "pigz ncbi_refseq_release_viral_plus_ttv_complete_genomes.fasta;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ps : Discuter et faire ici ajustement DB viral // TTV et aux seqs Orf1 et g_complet => OUI."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 5.1.2) Fungal DB"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "c ~/Desktop/TTV/BISCEm/Data/Fungal;\n",
    "BLAST='/home/erwann/Desktop/ncbi-blast-2.6.0+/bin';\n",
    "unpigz ncbi_refseq_release_fungi.fna.gz;\n",
    "$BLAST/makeblastdb -in ncbi_refseq_release_fungi.fna -input_type fasta -dbtype nucl -out rr_fungi;\n",
    "pigz ncbi_refseq_release_fungi.fna;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 5.1.3) Bacterial DB"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "blast='/home/erwann/Software/Ncbi_blast_2.6.0+/bin';\n",
    "unpigz hmp_bacterial.fasta.gz;\n",
    "$blast/makeblastdb -in hmp_bacterial.fasta -input_type fasta -dbtype nucl -out hmp_bacterial;\n",
    "pigz hmp_bacterial.fasta;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.2) Blastn vs viral DBs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Done locally :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "# Versus rr_viral_plus_ttv\n",
    "biscem='/home/erwann/Desktop/TTV/BISCEm';\n",
    "blast='/home/erwann/Software/Ncbi_blast_2.6.0+/bin';\n",
    "mkdir -p $biscem/Output/Viral_blastn;\n",
    "nt=8;\n",
    "\n",
    "cd $biscem/Output/Reads_qual_hm_fivep;\n",
    "for f in *.fasta;\n",
    "do $blast/blastn -db $biscem/Data/Blast_DB/rr_viral_plus_ttv \\\n",
    "                 -query $f \\\n",
    "                 -reward 1 \\\n",
    "                 -penalty -3 \\\n",
    "                 -word_size 12 \\\n",
    "                 -gapopen 5 \\\n",
    "                 -gapextend 2 \\\n",
    "                 -evalue 0.0001 \\\n",
    "                 -dust \"no\" \\\n",
    "                 -perc_identity 90 \\\n",
    "                 -culling_limit 2 \\\n",
    "                 -out $biscem/Output/Viral_blastn/${f%_*}\"_vs_viral.tsv\" \\\n",
    "                 -outfmt \"6 qseqid sseqid pident length mismatch gapopen qstart \\\n",
    "                          qend sstart send evalue bitscore qlen slen sstrand\" \\\n",
    "                 -num_threads $nt;\n",
    "done;\n",
    "\n",
    "# Versus rr_viral_plus_complete_genome\n",
    "biscem='/home/erwann/Desktop/TTV/BISCEm';\n",
    "blast='/home/erwann/Software/Ncbi_blast_2.6.0+/bin';\n",
    "mkdir -p $biscem/Output/Viral_blastn_complete;\n",
    "nt=8;\n",
    "\n",
    "cd $biscem/Output/Reads_qual_hm_fivep;\n",
    "for f in *.fasta;\n",
    "do $blast/blastn -db $biscem/Data/Blast_DB/rr_viral_plus_complete_genome \\\n",
    "                 -query $f \\\n",
    "                 -reward 1 \\\n",
    "                 -penalty -3 \\\n",
    "                 -word_size 12 \\\n",
    "                 -gapopen 5 \\\n",
    "                 -gapextend 2 \\\n",
    "                 -evalue 0.0001 \\\n",
    "                 -dust \"no\" \\\n",
    "                 -perc_identity 90 \\\n",
    "                 -culling_limit 2 \\\n",
    "                 -out $biscem/Output/Viral_blastn_complete/${f%_*}\"_vs_viral.tsv\" \\\n",
    "                 -outfmt \"6 qseqid sseqid pident length mismatch gapopen qstart \\\n",
    "                          qend sstart send evalue bitscore qlen slen sstrand\" \\\n",
    "                 -num_threads $nt;\n",
    "done;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.3) Blastn vs fungal DB"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Done on CALI :"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create an initial script :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "#!/bin/bash\n",
    "#SBATCH --partition=normal\n",
    "#SBATCH --ntasks=1\n",
    "#SBATCH --cpus-per-task=12\n",
    "#SBATCH --mem-per-cpu=8192\n",
    "#SBATCH --time=1-23:59:59\n",
    "#SBATCH --mail-user=erwan.scaon@unilim.fr\n",
    "#SBATCH --mail-type=begin\n",
    "#SBATCH --mail-type=end\n",
    "#SBATCH --error=/home/scaonp01/scratch/TTV/ttv_blastn_fungal_d5.verbose\n",
    "#SBATCH --output=/home/scaonp01/scratch/TTV/ttv_blastn_fungal_d5.verbose\n",
    "#SBATCH --job-name=ttv_blastn_fungal_d5\n",
    "\n",
    "# sbatch blastn_fungal_d5.sh\n",
    "\n",
    "##############################\n",
    "### Paths, tools & modules ###\n",
    "##############################\n",
    "nt=12;\n",
    "export OMP_NUM_THREADS=$SLURM_CPUS_PER_TASK;\n",
    "ttv='/home/scaonp01/TTV';\n",
    "scratch='/home/scaonp01/scratch/TTV';\n",
    "blast='/home/scaonp01/Softwares/Ncbi_blast_plus_2.6.0';\n",
    "\n",
    "################\n",
    "### Commands ###\n",
    "################\n",
    "cd $ttv/Reads_qual_hm_fivep;\n",
    "for f in d5_qual.fasta;\n",
    "do $blast/blastn -db $ttv/DB/rr_fungi \\\n",
    "                 -query $f \\\n",
    "                 -reward 1 \\\n",
    "                 -penalty -3 \\\n",
    "                 -word_size 12 \\\n",
    "                 -gapopen 5 \\\n",
    "                 -gapextend 2 \\\n",
    "                 -evalue 0.0001 \\\n",
    "                 -dust \"no\" \\\n",
    "                 -perc_identity 90 \\\n",
    "                 -culling_limit 2 \\\n",
    "                 -out $scratch/${f%_*}\"_vs_fungal.tsv\" \\\n",
    "                 -outfmt \"6 qseqid sseqid pident length mismatch gapopen \\\n",
    "                          qstart qend sstart send evalue bitscore qlen\" \\\n",
    "                 -num_threads $nt;\n",
    "done;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Launch script :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "sbatch blastn_fungal_d5.sh;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Duplicate it for the other samples :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "# Check what we are going to replace\n",
    "for id in 'd5';\n",
    "do grep --color -i $id blastn_fungal_\"$id\".sh;\n",
    "done;\n",
    "\n",
    "# Dup scripts\n",
    "for id in 'd1' 'd3' 'd4' 'd8' 'd10' 'd11' 'd12' 'd13' 'd14' 'd15' 'd16' 'd17' \\\n",
    "          'd19' 'd20' 'r1' 'r2' 'r3' 'r4' 'r5' 'r6' 'r7' 'r8' 'r9' 'r10' 'r11' \\\n",
    "          'r12' 'r13' 'r14' 'r15' 'r16' 'r17' 'r18' 'r19' 'r20' 'r21' 'r22' 'r23' \\\n",
    "          'r24' 'r25' 'r26' 'r27' 'r28' 'r29' 'r30' 'r31' 'r32' 'r33' 'r34' 'r35' \\\n",
    "          'r36' 'r37' 'r38' 'r39' 'r40' 'r41' 'r42' 'r43' 'r44' 'r45' 'r46';\n",
    "do cp blastn_fungal_d5.sh blastn_fungal_\"$id\".sh;\n",
    "done;\n",
    "\n",
    "# Replace with new E name\n",
    "for id in 'd1' 'd3' 'd4' 'd8' 'd10' 'd11' 'd12' 'd13' 'd14' 'd15' 'd16' 'd17' \\\n",
    "          'd19' 'd20' 'r1' 'r2' 'r3' 'r4' 'r5' 'r6' 'r7' 'r8' 'r9' 'r10' 'r11' \\\n",
    "          'r12' 'r13' 'r14' 'r15' 'r16' 'r17' 'r18' 'r19' 'r20' 'r21' 'r22' 'r23' \\\n",
    "          'r24' 'r25' 'r26' 'r27' 'r28' 'r29' 'r30' 'r31' 'r32' 'r33' 'r34' 'r35' \\\n",
    "          'r36' 'r37' 'r38' 'r39' 'r40' 'r41' 'r42' 'r43' 'r44' 'r45' 'r46';\n",
    "do sed -i \"s/d5/$id/\" blastn_fungal_\"$id\".sh;\n",
    "done;\n",
    "\n",
    "# Launch\n",
    "for id in 'd1' 'd3' 'd4' 'd8' 'd10' 'd11' 'd12' 'd13' 'd14' 'd15' 'd16' 'd17' \\\n",
    "          'd19' 'd20' 'r1' 'r2' 'r3' 'r4' 'r5' 'r6' 'r7' 'r8' 'r9' 'r10' 'r11' \\\n",
    "          'r12' 'r13' 'r14' 'r15' 'r16' 'r17' 'r18' 'r19' 'r20' 'r21' 'r22' 'r23' \\\n",
    "          'r24' 'r25' 'r26' 'r27' 'r28' 'r29' 'r30' 'r31' 'r32' 'r33' 'r34' 'r35' \\\n",
    "          'r36' 'r37' 'r38' 'r39' 'r40' 'r41' 'r42' 'r43' 'r44' 'r45' 'r46';\n",
    "do sbatch blastn_fungal_\"$id\".sh;\n",
    "done;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.4) Blastn vs bacterial DB"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Done on Genotoul :"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create an initial script (blastn_bacterial_d5.sh) :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### Code ###\n",
    "#!/bin/bash\n",
    "#$ -q workq\n",
    "#$ -M erwan.scaon@unilim.fr\n",
    "#$ -m bea\n",
    "#$ -l mem=8G\n",
    "#$ -l h_vmem=10G\n",
    "#$ -N TTV_blastn_bacterial_d5\n",
    "#$ -o /home/escaon/work/TTV/ttv_blastn_bacterial_d5_o.txt\n",
    "#$ -e /home/escaon/work/TTV/ttv_blastn_bacterial_d5_e.txt\n",
    "\n",
    "# qsub -pe parallel_smp $nt script.sh\n",
    "\n",
    "work='/home/escaon/work/TTV';\n",
    "save='/home/escaon/save/TTV';\n",
    "blast=\"$work/NCBI_BlastPlus_2.6.0/bin\";\n",
    "mkdir -p $work/Bacterial_blastn;\n",
    "nt=8;\n",
    "\n",
    "cd $save/Reads_qual_hm_fivep;\n",
    "for f in d5_qual.fasta;\n",
    "do $blast/blastn -db $work/Blast_db/hmp_bacterial \\\n",
    "                 -query $f \\\n",
    "                 -reward 1 \\\n",
    "                 -penalty -3 \\\n",
    "                 -word_size 12 \\\n",
    "                 -gapopen 5 \\\n",
    "                 -gapextend 2 \\\n",
    "                 -evalue 0.0001 \\\n",
    "                 -dust \"no\" \\\n",
    "                 -perc_identity 90 \\\n",
    "                 -culling_limit 2 \\\n",
    "                 -out $work/Bacterial_blastn/${f%_*}\"_vs_bacterial.tsv\" \\\n",
    "                 -outfmt \"6 qseqid sseqid pident length mismatch gapopen \\\n",
    "                          qstart qend sstart send evalue bitscore qlen\" \\\n",
    "                 -num_threads $nt;\n",
    "done;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Launch it :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "qsub -pe parallel_smp 8 blastn_bacterial_d5.sh;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Duplicate it for the other samples :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#### Code ###\n",
    "# Check what we are going to replace\n",
    "for id in 'd5';\n",
    "do grep --color -i $id blastn_bacterial_\"$id\".sh;\n",
    "done;\n",
    "\n",
    "# Dup scripts\n",
    "for id in 'd1' 'd3' 'd4' 'd8' 'd10' 'd11' 'd12' 'd13' 'd14' 'd15' 'd16' 'd17' \\\n",
    "          'd19' 'd20' 'r1' 'r2' 'r3' 'r4' 'r5' 'r6' 'r7' 'r8' 'r9' 'r10' 'r11' \\\n",
    "          'r12' 'r13' 'r14' 'r15' 'r16' 'r17' 'r18' 'r19' 'r20' 'r21' 'r22' 'r23' \\\n",
    "          'r24' 'r25' 'r26' 'r27' 'r28' 'r29' 'r30' 'r31' 'r32' 'r33' 'r34' 'r35' \\\n",
    "          'r36' 'r37' 'r38' 'r39' 'r40' 'r41' 'r42' 'r43' 'r44' 'r45' 'r46';\n",
    "do cp blastn_bacterial_d5.sh blastn_bacterial_\"$id\".sh;\n",
    "done;\n",
    "\n",
    "# Replace with new E name\n",
    "for id in 'd1' 'd3' 'd4' 'd8' 'd10' 'd11' 'd12' 'd13' 'd14' 'd15' 'd16' 'd17' \\\n",
    "          'd19' 'd20' 'r1' 'r2' 'r3' 'r4' 'r5' 'r6' 'r7' 'r8' 'r9' 'r10' 'r11' \\\n",
    "          'r12' 'r13' 'r14' 'r15' 'r16' 'r17' 'r18' 'r19' 'r20' 'r21' 'r22' 'r23' \\\n",
    "          'r24' 'r25' 'r26' 'r27' 'r28' 'r29' 'r30' 'r31' 'r32' 'r33' 'r34' 'r35' \\\n",
    "          'r36' 'r37' 'r38' 'r39' 'r40' 'r41' 'r42' 'r43' 'r44' 'r45' 'r46';\n",
    "do sed -i \"s/d5/$id/\" blastn_bacterial_\"$id\".sh;\n",
    "done;\n",
    "\n",
    "# Launch\n",
    "for id in 'd1' 'd3' 'd4' 'd8' 'd10' 'd11' 'd12' 'd13' 'd14' 'd15' 'd16' 'd17' \\\n",
    "          'd19' 'd20' 'r1' 'r2' 'r3' 'r4' 'r5' 'r6' 'r7' 'r8' 'r9' 'r10' 'r11' \\\n",
    "          'r12' 'r13' 'r14' 'r15' 'r16' 'r17' 'r18' 'r19' 'r20' 'r21' 'r22' 'r23' \\\n",
    "          'r24' 'r25' 'r26' 'r27' 'r28' 'r29' 'r30' 'r31' 'r32' 'r33' 'r34' 'r35' \\\n",
    "          'r36' 'r37' 'r38' 'r39' 'r40' 'r41' 'r42' 'r43' 'r44' 'r45' 'r46';\n",
    "do qsub -pe parallel_smp 8 blastn_bacterial_\"$id\".sh;\n",
    "done;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.5) Hit count for DBs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "##### Viral DB :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "cd ~/Desktop/TTV/BISCEm/Output/Viral_blastn;\n",
    "# Discard blastn hits shorter than 90% of query(read)_len\n",
    "for f in *.tsv;\n",
    "do awk -F \"\\t\" '$4>=($13*90/100) {print $0}' $f \\\n",
    "       > ${f%%_*}\"_vs_viral_threshed.tsv\";\n",
    "done;\n",
    "# Print unique reads ids (rid) with at least 1 hit in each TSV\n",
    "for f in *_threshed.tsv;\n",
    "do printf $f\"\\t\";\n",
    "   awk '{print $1}' $f \\\n",
    "       | sort -u \\\n",
    "       | wc -l;\n",
    "done;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Per TTV genotypes Orf1 inside Viral DB :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "for f in *_threshed.tsv\n",
    "python ./counts_orf1.py "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "##### Fungal :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "##### Bacterial :"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Retrieve blastn results from genotoul :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "# some verifications first\n",
    "grep -i error *.txt # OK\n",
    "l -1 Bacterial_blastn/*.tsv | wc -l # 61\n",
    "# & check that all mails have \"Exit Status = 0\"\n",
    "# => OK\n",
    "\n",
    "# DL locally\n",
    "sudo scp -r escaon@genotoul.toulouse.inra.fr:/home/escaon/work/TTV/Bacterial_blastn \\\n",
    "/home/erwann/Desktop/TTV/BISCEm/Output;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Get number of uniques reads matching bacterial DB :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "cd ~/Desktop/TTV/BISCEm/Output/Bacterial_blastn;\n",
    "\n",
    "# Discard blastn hits shorter than 90% of query(read)_len\n",
    "for f in *.tsv;\n",
    "do awk -F \"\\t\" '$4>=($13*90/100) {print $0}' $f \\\n",
    "       > ${f%%_*}\"_vs_bacterial_threshed.tsv\";\n",
    "done;\n",
    "\n",
    "# Retrieve unique reads ids (rid) with at least 1 hit in each TSV\n",
    "for f in *_threshed.tsv;\n",
    "do awk '{print $1}' $f \\\n",
    "       | sort -u \\\n",
    "       > ${f%%_*}_threshed_rid.txt;\n",
    "done;\n",
    "wc -l *.txt;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.6) Hit count per TTV genotypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "code_folding": [],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### CODE ###\n",
    "# Orf1\n",
    "cd ~/Desktop/TTV/BISCEm/Output/Viral_blastn;\n",
    "for f in *_threshed.tsv;\n",
    "do mkdir -p ${f%%_*};\n",
    "   cd ${f%%_*};\n",
    "   python ../../../Scripts/counts_orf1.py ../$f;\n",
    "   cd ..;\n",
    "done;\n",
    "\n",
    "# Complete\n",
    "cd ~/Desktop/TTV/BISCEm/Output/Viral_blastn_complete;\n",
    "for f in *_threshed.tsv;\n",
    "do python ../../Scripts/counts_per_g_complete.py $f;\n",
    "done;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## X) Bulk ideas / TODO"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Do plots (camemberts avec != cutoffs, 100, 1000...) <br>\n",
    "- Brainstorming about assembly step. <br>\n",
    "- Tools to try : <br>\n",
    "https://sourceforge.net/projects/sb2nhri/files/drVM/ <br>\n",
    "http://ivory.idyll.org/blog/2017-classify-genome-bins-with-custom-db-try-again.html <br>\n",
    "https://www.biorxiv.org/content/early/2017/10/02/196998 <br>\n",
    "\n",
    "\n",
    "Obtenir un compte non-biaisé du nombre de reads par génotypes de TTV.\n",
    "\n",
    "Estimer le pourcentage de reads humains, bactériens, fongiques et viraux.\n",
    "\n",
    "Identifier la ou les souches de TTV qui sont transmises du donneur au receveur et la ou les souches de TTV qui expliquent la majeur partie de la charge virale.\n",
    "\n",
    "Qui des \"multi best-hits\" <=> sur 2 génome différents avec de bon scores ? -> Discard ?\n",
    "\n",
    "Puce 1 = Dorian = Manuel, Puce 2 = Dorian = library-builder, Puce 3 = GenoLim, 3.33 = Manuel, 3.63 = library-builder.\n",
    "\n",
    "Le biais de hit envers Orf1 s'expliquerai-t-il par multi best-hit d'un même hit sur Orf1 et sur une autre zone chez les TTV (en supposant que ce soit majoritairement l'autre zone qui ait un meilleur score et qui soit donc sélectionnée). Ou alors pour une raison bio-mol qui m'échappe, plus faible qualité des reads qui span majoritairement sur Orf1, du coup ils auraient en grande partie sauté lors du trimming ?\n",
    "\n",
    "L'estimation de la charge virale a été faite avant le rolling circle (nous avons les chiffres, estimée via les 128 nts en 5' via primer de Dorian) et après (manque les chiffres) -> need them.\n",
    "\n",
    "Cinétique des G (poster vu par Sophie) -> % de chaque G // charge vir ? G et apparition/disparition.\n",
    "\n",
    "Plots : Faire un volet génomes complets et en regard un volet avec les autres génomes.\n",
    "\n",
    "Filter : Etape actuelles vs vir_DB ne serviraient qu'a choper des reads candidats qu'on re-blasterai avec des params bcp + spécifiques contre les TTV ensuite ?\n",
    "\n",
    "X) L'analyse comportera 2 \"embranchements\" : <br>\n",
    "- Chemin Orf1 || génomes_complets\n",
    "- Puis raisonnement sur génotypes pris 1 à 1 || groupes phylo de génotypes. <br>\n",
    "Ps : Pour voir si on s'attend à une grande différence entre Orf1 & complet, plot par génotypes le pileup des reads (des hits) sur séquences génome complet de l'étude. <br>\n",
    "\n",
    "A) Décrire les génotypes de TTV présents chez les D & R (D_all, R_all, R_35, R_75, R_360). <br>\n",
    "\n",
    "B) Des génotypes de TTV sont-ils transmis de D vers R, si oui quels sont-ils (Ps : Sachant que la charge virale est faible ches les D, si un génotype est présent chez D, alors on doit le rechercher chez tous les R) ? Certains ne sont-ils jamais transmis ? Taux de transmission ? Taux / succès de transmission est-il lié à la charge virale du D ? <br>\n",
    "\n",
    "C) Quels sont les génotypes qui croissent ou décroissent chez les R (quel(s) G(s) explique(nt) la plus grande partie de l'augmentation de la charge virale) ? Viennent-ils de D ? Sont-ce les mêmes 2/3 génotypes / groupe phylo de génotypes qui croissent chez tout les R ? (doit-on rapporter à la charge virale ?) <br>\n",
    "\n",
    "D) Les génotypes trouvés chez D & R sont-ils les mêmes d'un point de vue séquence ? Ex : D transmet G18 à R. G18 subit-il des modifs chez R ? (need assembly or something here => tool publi gigascience ?). Ps : Check aussi si recombinaison entre Gx & Gy (check via reads chevauchants ?). Un point important ici va être de donner la bonne sélection de reads en entrée d'un éventuel assemblage (donner filtered -Bact -Fungus ? Quid de refiltrer sur les no-match?). Et selon ce qu'on arrive à assembler, il faut prendre à peu près la même séquence pour comparer ensuite (genre si on assemble 3 G complet et 7 Orf1, prendre 10 Orf1 pour les comparaisons). <br>\n",
    "\n",
    "Exploratoire) \"Top 10\" des familles / espèces / ... les plus fréquentes lors du blastn vs Viral, Fungal & Bacterial. Pour max X reads qui ne se sont alignés contre rien, blast vs nt / nr."
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Bash",
   "language": "bash",
   "name": "bash"
  },
  "language_info": {
   "codemirror_mode": "shell",
   "file_extension": ".sh",
   "mimetype": "text/x-sh",
   "name": "bash"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}