annotate test-data/test/small_rna_map.bash @ 5:d65045e976e6 draft

planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit b673d39fbe79f5164ba6489b33cfa78ac238ee09
author artbio
date Sat, 22 Jul 2017 11:45:52 -0400
parents 1ad5d040f85f
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
1 #! /bin/bash
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
2 # Execution command : bash small_rna_map.bash <sam file>
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
3 samtools view -h -o input.sam ../test-data/input.bam
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
4 samfile="input.sam"
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
5 cut -f2,3,4,5,6 ../test-data/output.tab > test-data.dat
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
6 grep -e "^@SQ" $samfile > header.tab
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
7 cat $samfile | perl -ne 'print unless /^@/ or ( (split /\t/)[1] != 0 and (split /\t/)[1] != 16 )' > output1.tab
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
8
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
9 cat << EOF > pyscript.py
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
10 lendic = {}
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
11 for line in open("header.tab"):
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
12 fields = line[:-1].split()
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
13 lendic[fields[1][3:]] = fields[2][3:]
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
14 readdic = {}
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
15 for line in (open("output1.tab")):
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
16 fields = line[:-1].split()
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
17 key = "-".join([fields[2],fields[3],"F" if fields[1]=="0" else "R"])
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
18 try:
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
19 readdic[key] += 1
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
20 except KeyError:
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
21 readdic[key] = 1
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
22 Out = open("output.tab", "w")
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
23 Out.write("Chromosome\tChrom_length\tCoordinate\tNbr_reads\tPolarity\n")
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
24 for key in sorted(readdic):
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
25 fields = key.split("-")
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
26 Chromosome, Coordinate, Polarity = fields[0], fields[1], fields[2]
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
27 Chrom_length = lendic[Chromosome]
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
28 Nbr_reads = str(readdic[key])
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
29 Out.write("\t".join([Chromosome, Chrom_length, Coordinate, Nbr_reads, Polarity]) )
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
30 Out.write("\n")
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
31 Out.close
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
32 EOF
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
33
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
34 python ./pyscript.py
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
35 rm output1.tab header.tab pyscript.py
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
36 sort -k 1,1 -k 3,3n -k 5,5 output.tab -o output.tab
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
37 if ! diff -q output.tab test-data.dat &>/dev/null; then
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
38 >&2 echo "different"
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
39 else
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
40 >&2 echo "Test passed"
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
41 fi
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
42
1ad5d040f85f planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_map commit fa452d860cf550c7524df59e77b36fd39e3e2a45
artbio
parents:
diff changeset
43 rm output.tab input.sam test-data.dat