Mercurial > repos > davidvanzessen > mutation_analysis

comparison wrapper.sh @ 19:c518cf0d4adb draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author davidvanzessen
date Wed, 01 Apr 2015 05:09:59 -0400
parents 2bb24ebbbc38
children 850857bc8605
comparison
equal deleted inserted replaced
18:9dd557f17153 19:c518cf0d4adb
1 #!/bin/bash 1 #!/bin/bash
2 set -e 2 set -e
3 dir="$(cd "$(dirname "$0")" && pwd)" 3 dir="$(cd "$(dirname "$0")" && pwd)"
4 input=$1 4 input=$1
5 output=$2 5 method=$2
6 outdir=$3 6 output=$3
7 title=$4 7 outdir=$4
8 title=$5
8 mkdir $outdir 9 mkdir $outdir
9 10
10 unzip $input -d $PWD/files/ > $PWD/unziplog.log 11 unzip $input -d $PWD/files/ > $PWD/unziplog.log
11 cat $PWD/files/*/1_* > $PWD/summary.txt 12 cat $PWD/files/*/1_* > $PWD/summary.txt
12 cat $PWD/files/*/7_* > $PWD/mutationanalysis.txt 13 cat $PWD/files/*/7_* > $PWD/mutationanalysis.txt
13 cat $PWD/files/*/8_* > $PWD/mutationstats.txt 14 cat $PWD/files/*/8_* > $PWD/mutationstats.txt
14 cat $PWD/files/*/10_* > $PWD/hotspots.txt 15 cat $PWD/files/*/10_* > $PWD/hotspots.txt
15 16
16 echo "${BLASTN}" 17 BLASTN_DIR="/home/galaxy/tmp/blast/ncbi-blast-2.2.30+/bin"
18
19 echo "${BLASTN_DIR}"
17 20
18 21
19 echo "identification" 22
20 python $dir/gene_identification.py --input $PWD/summary.txt --output $outdir/annotatedsummary.txt 23
24 echo "identification ($method)"
25
26 if [[ "${method}" == "custom" ]] ; then
27 python $dir/gene_identification.py --input $PWD/summary.txt --output $outdir/identified_genes.txt
28 else
29 ID_index=$(cat $PWD/summary.txt | grep -o -P ".+Sequence ID" | grep -o -P "\t" | wc -l)
30 ID_index=$((ID_index+1))
31 sequence_index=$(cat $PWD/summary.txt | grep -o -P ".+\tSequence" | grep -o -P "\t" | wc -l)
32 sequence_index=$((sequence_index+1))
33
34 echo "$ID_index ${sequence_index}"
35
36 cat $PWD/summary.txt | tail -n+2 | cut -f ${ID_index},${sequence_index} | awk '{print ">" $1 "\n" $2}' > $PWD/sequences.fasta
37
38 echo -e "qseqid\tsseqid\tpident\tlength\tmismatch\tgapopen\tqstart\tqend\tsstart\tsend\tevalue\tbitscore" > $outdir/identified_genes.txt
39 ${BLASTN_DIR}/blastn -task blastn -db $dir/subclass_definition.db -query $PWD/sequences.fasta -outfmt 6 >> $outdir/identified_genes.txt
40 fi
41
42
43
21 echo "merging" 44 echo "merging"
22 Rscript $dir/merge_and_filter.r $outdir/annotatedsummary.txt $PWD/mutationanalysis.txt $PWD/mutationstats.txt $PWD/hotspots.txt $outdir/merged.txt $outdir/unmatched.txt 45 Rscript $dir/merge_and_filter.r $PWD/summary.txt $PWD/mutationanalysis.txt $PWD/mutationstats.txt $PWD/hotspots.txt $outdir/identified_genes.txt $outdir/merged.txt $outdir/unmatched.txt $method
23 46
24 genes="ca,ca1,ca2,cg,cg1,cg2,cg3,cg4,cm" 47 genes="ca,ca1,ca2,cg,cg1,cg2,cg3,cg4,cm"
25 echo "R mutation analysis" 48 echo "R mutation analysis"
26 Rscript $dir/mutation_analysis.r $outdir/merged.txt $genes $outdir 2>&1 49 Rscript $dir/mutation_analysis.r $outdir/merged.txt $genes $outdir 2>&1
27 echo "python mutation analysis" 50 echo "python mutation analysis"