Mercurial > repos > davidvanzessen > sff_extract_demultiplex
comparison wrapper.sh @ 0:cb08a27e5fc2 draft
Uploaded
| author | davidvanzessen |
|---|---|
| date | Mon, 29 Aug 2016 05:44:57 -0400 |
| parents | |
| children | 44ec2a1009fc |
comparison
equal
deleted
inserted
replaced
| -1:000000000000 | 0:cb08a27e5fc2 |
|---|---|
| 1 #!/bin/bash | |
| 2 input="$1" | |
| 3 output="$2" | |
| 4 outDir="$3" | |
| 5 mkdir "$outDir" | |
| 6 EOL="$4" | |
| 7 mismatches="$5" | |
| 8 partial="$6" | |
| 9 name=$(basename "$7") | |
| 10 ext="${name##*.}" | |
| 11 name="${name%.*}" | |
| 12 name="${name// /_}" | |
| 13 prefix="${name}_" | |
| 14 dir="$(cd "$(dirname "$0")" && pwd)" | |
| 15 | |
| 16 unzip $dir/fastqc_v0.11.2.zip -d $PWD/ > $PWD/unziplog.log | |
| 17 chmod 755 $PWD/FastQC/fastqc | |
| 18 | |
| 19 declare -A trim_start | |
| 20 declare -A trim_end | |
| 21 for ((i=8;i<=$#;i=i+4)) | |
| 22 do | |
| 23 j=$((i+1)) | |
| 24 start_int=$((i+2)) | |
| 25 end_int=$((i+3)) | |
| 26 id="${!i}" | |
| 27 echo "$id, ${start_int}, ${end_int}" | |
| 28 trim_start[$id]=${!start_int} | |
| 29 trim_end[$id]=${!end_int} | |
| 30 echo -e "$id\t${!j}" >> $outDir/barcodes.txt | |
| 31 | |
| 32 done | |
| 33 trim_start["unmatched"]=0 | |
| 34 trim_end["unmatched"]=0 | |
| 35 | |
| 36 echo "trim_start = ${trim_start[@]}" | |
| 37 echo "trim_end = ${trim_end[@]}" | |
| 38 | |
| 39 workdir=$PWD | |
| 40 cd $outDir | |
| 41 echo "$3" | |
| 42 filetype=`file $input` | |
| 43 result="" | |
| 44 if [[ $filetype == *ASCII* ]] | |
| 45 then | |
| 46 result=`cat $input | $dir/fastx_barcode_splitter.pl --bcfile $outDir/barcodes.txt --prefix "$prefix" --suffix ".fastq" --$EOL --mismatches $mismatches --partial $partial` | |
| 47 else | |
| 48 result=`$dir/sff2fastq $input | $dir/fastx_barcode_splitter.pl --bcfile $outDir/barcodes.txt --prefix "$prefix" --suffix ".fastq" --$EOL --mismatches $mismatches --partial $partial` | |
| 49 fi | |
| 50 | |
| 51 echo "$result" | tail -n +2 | sed 's/\t/,/g' > output.txt | |
| 52 echo "<html><head><title>$name demultiplex</title></head><body><table border='1'><thead><tr><th>ID</th><th>Count</th><th>FASTQ</th><th>FASTA</th><th>Trimmed FASTA</th><th>FASTQC</th></tr></thead><tbody>" >> $output | |
| 53 while IFS=, read barcode count location | |
| 54 do | |
| 55 if [ "total" == "$barcode" ] | |
| 56 then | |
| 57 echo "<tr><td>$barcode</td><td>$count</td><td></td><td></td><td></td><td></td><td></td><td></td></tr>" >> $output | |
| 58 break | |
| 59 fi | |
| 60 file="${name}_${barcode}" | |
| 61 mkdir "$outDir/fastqc_$barcode" | |
| 62 $workdir/FastQC/fastqc "$file.fastq" -o "$outDir" 2> /dev/null | |
| 63 cat "$file.fastq" | awk 'NR%4==1{printf ">%s\n", substr($0,2)}NR%4==2{print}' > "$file.fasta" | |
| 64 python $dir/trim.py --input "$file.fasta" --output "${file}_trimmed.fasta" --start "${trim_start[$barcode]}" --end "${trim_end[$barcode]}" | |
| 65 echo "<tr><td>$barcode</td><td>$count</td><td><a href='$file.fastq'>$file.fastq</a></td><td><a href='$file.fasta'>$file.fasta</a></td><td><a href='${file}_trimmed.fasta'>${file}_trimmed.fasta</a></td><td><a href='${name}_${barcode}_fastqc.html'>Report</a></td></tr>" >> $output | |
| 66 done < output.txt | |
| 67 echo "</tbody></body></html>" >> $output |