{
 "metadata": {
  "name": "",
  "signature": "sha256:761d87985a3f607000f91c3913b2df44bbbdf8b01a5e554844bd27f354c42461"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Post Manuscript Review - data crunch"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#file ID\n",
      "#fid=\"CgM1\"\n",
      "\n",
      "#TIMESTAMP\n",
      "date=!date +%m%d_%H%M\n",
      "\n",
      "#working directory (parent)\n",
      "#wd=\"/Volumes/web/cnidarian/BiGo_larvae_merge/\"\n",
      "\n",
      "#where is bsmap\n",
      "#bsmap=\"/Users/Shared/Apps/bsmap-2.73/\"\n",
      "bsmap=\"/Volumes/Bay3/Software/BSMAP/bsmap-2.74/\"\n",
      "\n",
      "#fastq files location R1 location\n",
      "R1=\"/Volumes/web/trilobite/Crassostrea_gigas_HTSdata/filtered_174gm_A_BiGosperm_L006_R1.fastq\"\n",
      "\n",
      "#fastq files location R2 location\n",
      "#comment out if SE\n",
      "R2=\"/Volumes/web/trilobite/Crassostrea_gigas_HTSdata/filtered_174gm_A_BiGosperm_L006_R2.fastq\"\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 3
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#option - number of processes \n",
      "!{bsmap}bsmap -a {R1} -b {R2} -d /Volumes/web/cnidarian/oyster.v9.fa -o /Volumes/web/cnidarian/BiGo_bsmap_v9_{date}.sam -p 3"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "###bsmap on iplant\n",
      "```\n",
      "Start at:  Fri Apr  4 06:42:38 2014\n",
      "\n",
      "Input reference file: oyster.v9.fa.gz \t(format: gzipped FASTA)\n",
      "Load in 11969 db seqs, total size 558601156 bp. 13 secs passed\n",
      "total_kmers: 43046721\n",
      "Create seed table. 36 secs passed\n",
      "max number of mismatches: read_length * 8% \tmax gap size: 0\n",
      "kmer cut-off ratio: 5e-07\n",
      "max multi-hits: 100\tmax Ns: 5\tseed size: 16\tindex interval: 4\n",
      "quality cutoff: 0\tbase quality char: '!'\n",
      "min fragment size:28\tmax fragemt size:500\n",
      "start from read #1\tend at read #4294967295\n",
      "additional alignment: T in reads => C in reference\n",
      "mapping strand (read_1): ++,-+\n",
      "mapping strand (read_2): +-,--\n",
      "Pair-end alignment(8 threads)\n",
      "Input read file #1: filtered_174gm_A_BiGosperm_L006_R1.fastq \t(format: FASTQ)\n",
      "Input read file #2: filtered_174gm_A_BiGosperm_L006_R2.fastq \t(format: FASTQ)\n",
      "Output file: bsmap_out.sam\t (format: SAM)\n",
      "Thread #2: \t100000 read pairs finished. 52 secs passed\n",
      "Total number of aligned reads: \n",
      "pairs:       90102067 (53%)\n",
      "single a:    17033002 (9.9%)\n",
      "single b:    15975991 (9.3%)\n",
      "Done.\n",
      "Finished at Fri Apr  4 09:02:07 2014\n",
      "Total time consumed:  8369 secs\n",
      "\n",
      "```"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "!curl -o 'BiGo_sperm_v9bsmap_out.sam' 'http://de.iplantcollaborative.org/dl/d/AE429A54-2B38-44B0-8AD7-E92DCDA2DEE8/bsmap_out.sam'"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current\r\n",
        "                                 Dload  Upload   Total   Spent    Left  Speed\r\n",
        "\r",
        "  0     0    0     0    0     0      0      0 --:--:-- --:--:-- --:--:--     0"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100  1149    0  1149    0     0    937      0 --:--:--  0:00:01 --:--:--  1354"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100  637k    0  637k    0     0   334k      0 --:--:--  0:00:01 --:--:--  417k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 2104k    0 2104k    0     0   722k      0 --:--:--  0:00:02 --:--:--  831k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 3610k    0 3610k    0     0   921k      0 --:--:--  0:00:03 --:--:-- 1019k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 4990k    0 4990k    0     0  1013k      0 --:--:--  0:00:04 --:--:-- 1098k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 6488k    0 6488k    0     0  1094k      0 --:--:--  0:00:05 --:--:-- 1379k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 7934k    0 7934k    0     0  1145k      0 --:--:--  0:00:06 --:--:-- 1453k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 9352k    0 9352k    0     0  1179k      0 --:--:--  0:00:07 --:--:-- 1444k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 10.6M    0 10.6M    0     0  1218k      0 --:--:--  0:00:08 --:--:-- 1452k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 12.0M    0 12.0M    0     0  1244k      0 --:--:--  0:00:09 --:--:-- 1471k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 13.5M    0 13.5M    0     0  1268k      0 --:--:--  0:00:10 --:--:-- 1474k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 14.8M    0 14.8M    0     0  1277k      0 --:--:--  0:00:11 --:--:-- 1462k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 16.1M    0 16.1M    0     0  1285k "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "     0 --:--:--  0:00:12 --:--:-- 1453k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 17.1M    0 17.1M    0     0  1266k      0 --:--:--  0:00:13 --:--:-- 1351k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 18.3M    0 18.3M    0     0  1259k      0 --:--:--  0:00:14 --:--:-- 1289k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 19.6M    0 19.6M    0     0  1264k      0 --:--:--  0:00:15 --:--:-- 1256k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 20.8M    0 20.8M    0     0  1264k      0 --:--:--  0:00:16 --:--:-- 1232k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 22.0M    0 22.0M    0     0  1262k      0 --:--:--  0:00:17 --:--:-- 1205k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 23.4M    0 23.4M    0     0  1268k      0 --:--:--  0:00:18 --:--:-- 1273k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 24.7M    0 24.7M    0     0  1274k      0 --:--:--  0:00:19 --:--:-- 1319k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 26.2M    0 26.2M    0     0  1286k      0 --:--:--  0:00:20 --:--:-- 1358k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 27.6M    0 27.6M    0     0  1293k      0 --:--:--  0:00:21 --:--:-- 1390k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 29.0M    0 29.0M    0     0  1298k      0 --:--:--  0:00:22 --:--:-- 1426k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 30.5M    0 30.5M    0     0  1307k      0 --:--:--  0:00:23 --:--:-- 1453k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 31.8M    0 31.8M    0     0  1309k      0 --:--:--  0:00:24 --:--:-- 1450k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 33.1M    0 33.1M    0     0  1310k      0 --:--:--  0:00:25 --:--:-- 1410k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 34.3M    0 34.3M    0     0  1306k      0 --:--:--  0:00:26 --:--:-- 1367k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 35.5M    0 35.5M    0     0  1303k      0 --:--:--  0:00:27 --:--:-- 1326k"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        "100 36.7M    0 36.7M    0     0  1301k      0 --:--:--  0:00:28 --:--:-- 1275k"
       ]
      }
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "running fastqc on iplant"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "!python {bsmap}methratio.py -d {genome} -u -z -g -o methratio_out.txt -s {bsmap}samtools bsmap_out.sam \n",
      "#command for only obtaining the context '__CG_'\n",
      "!grep \"[A-Z][A-Z]CG[A-Z]\" <methratio_out.txt> methratio_out_CG.txt \n",
      "#5x coverage\n",
      "!awk '{if ($8 >= 5) print $1,$2-1,$2+1,\"CpG\",$5}' <methratio_out_CG.txt> filt_methratio_out_CG.igv \n",
      "!tr ' ' \"\\t\" <filt_methratio_out_CG.igv> filt_methratio_{fid}.igv"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [],
     "language": "python",
     "metadata": {},
     "outputs": []
    }
   ],
   "metadata": {}
  }
 ]
}