rollback encode changes in add user branch

#!/bin/bash

# Converts bam file to merged_nodups.txt file

# Calls samtools to filter duplicates then 

# chimeric_blacklist to associate read pair group

# Then assigns fragment and sorts

# Sorting should be optional

# Usage: bamtotxt.sh <bam_file> <site_file> <output_file> [sort]

set -e

if [ $# -lt 3 ] || [ $# -gt 4 ]

then

    echo "Usage: $0 <bam_file> <site_file> <output_file> [sort]"

    exit 1

else

    bamfile=$1

    sitefile=$2 

    outputfile=$3

fi

if [ $# -eq 4 ]

then

    sort=1

fi

samtools view -F 1024 "$bamfile" | awk -v fname=$bamfile -f "$( dirname "${BASH_SOURCE[0]}" )"/chimeric_blacklist.awk

"$( dirname "${BASH_SOURCE[0]}" )"/fragment.pl "${bamfile}_norm.txt" "${bamfile}.frag.txt" "$sitefile"

if [ -n $sort ]

then

    sort -k2,2d -k6,6d -k4,4n -k8,8n -k1,1n -k5,5n -k3,3n "${bamfile}.frag.txt" > "$outputfile"

else

    mv "${bamfile}.frag.txt" "$outputfile"

fi

rm "${bamfile}_norm.txt" "${bamfile}.frag.txt"

# Start by checking the statistics to see if they add up 

res1=`cat ${splitdir}/*norm*res*`

check1=`cat ${splitdir}/*norm*res* | awk '{s2+=$2; s3+=$3; s4+=$4; s5+=$5; s6+=$6; s7+=$7; s8+=$8}END{if (s2 != s3+s4+s5+s6+s7+s8){print 0}else{print 1}}'`

check1=`cat ${splitdir}/*norm*res* | awk '{s2+=$2; s3+=$3; s4+=$4; s5+=$5; s6+=$6}END{if (s2 != s3+s4+s5+s6){print 0}else{print 1}}'`

if [ $check1 -eq 0 ] || [ -z "$res1" ]

then

    echo "***! Error! The statistics do not add up. Alignment likely failed to complete on one or more files. Run relaunch_prep.sh"

# Script to clean up big repetitive files and zip fastqs. Run after you are 

# sure the pipeline ran successfully.  Run from top directory (HIC001 e.g.).

# Juicer version 1.5

total=`ls -l aligned/merged_sort.txt | awk '{print $5}'`

total2=`ls -l aligned/merged_nodups.txt aligned/dups.txt aligned/opt_dups.txt | awk '{sum = sum + $5}END{print sum}'`

if [ $total -eq $total2 ] 

then 

    rm aligned/merged_sort.txt 

    rm -rf splits 

    testname=$(ls -l fastq | awk 'NR==1{print $9}')

    if [ "${testname: -5}" == ".fastq" ]

    then

    rm -r splits 

    for i in fastq/*.fastq

    do

        gzip $i

    done

    fi

    gzip aligned/merged_nodups.txt

    gzip aligned/dups.txt

    gzip aligned/opt_dups.txt

    gzip aligned/abnormal.sam

    gzip aligned/collisions.txt

    gzip aligned/unmapped.sam

else 

    echo "Problem: The sum of merged_nodups and the dups files is not the same size as merged_sort.txt"

    echo "Did NOT clean up";

# Juicer version 1.5

if [ "$usegzip" -eq 1 ]

then 

    num1=$(paste <(gunzip -c $file1) <(gunzip -c $file2) | grep -cE $ligation)

    num2=$(gunzip -c $file1 | wc -l | awk '{print $1}')

    num1=$(paste <(gunzip -c $name1$ext) <(gunzip -c $name2$ext) | grep -cE $ligation)

    num2=$(gunzip -c ${name1}${ext} | wc -l | awk '{print $1}')

else

    num1=$(paste $file1 $file2 | grep -cE $ligation)

    num2=$(wc -l $file1 | awk '{print $1}')

    num1=$(paste $name1$ext $name2$ext | grep -cE $ligation)

    num2=$(wc -l ${name1}${ext} | awk '{print $1}')

fi

echo -ne "$num1 " > ${curr_ostem}_norm.txt.res.txt

echo "$num2" > ${curr_ostem}_linecount.txt

echo -ne "$num1 " > ${name}${ext}_norm.txt.res.txt

echo "$num2" > ${name}${ext}_linecount.txt