def _hisat(input, index, mirbase):
safe_makedir("hisat")
with ch_directory("hisat"):
output = "hisat_map.bam"
cmd = ("hisat -f -k 50 -L 18 -x {index}" " -U {input}" " >| hits.sam")
cmd_bam = "samtools view -Sbh hits.sam >| {output}"
if not is_there("hits.sam"):
do.run(cmd.format(**locals()), "")
if not is_there(output):
do.run(cmd_bam.format(**locals()), "")
mirbase_output = _annotate(output, mirbase)
return (mirbase_output, full(output))
def _tailor(input, index, mirbase):
safe_makedir("tailor")
with ch_directory("tailor"):
output = "tailor_map.bam"
cmd = ("tailor -l 15 map -p {index}" " -i {input}" " -o hits.sam")
cmd_bam = "samtools view -Sbh hits.sam >| {output}"
if not is_there("hits.sam"):
do.run(cmd.format(**locals()), "")
if not is_there(output):
do.run(cmd_bam.format(**locals()), "")
mirbase_output = _annotate(output, mirbase)
return (mirbase_output, full(output))
def _hisat(input, index, mirbase):
with ch_directory("hisat"):
output = "hisat_map.bam"
cmd = ("hisat -k 50 -L 18 -x {index}"
" -U {input}"
" >| hits.sam")
cmd_bam = "samtools view -Sbh hits.sam >| {output}"
if not is_there("hits.sam"):
do.run(cmd.format(**locals()), "")
if not is_there(output):
do.run(cmd_bam.format(**locals()), "")
mirbase_output = _annotate(output, mirbase)
return mirbase_output
def _tailor(input, index, mirbase):
safe_makedir("tailor")
with ch_directory("tailor"):
output = "tailor_map.bam"
cmd = ("tailor -l 15 map -p {index}"
" -i {input}"
" -o hits.sam")
cmd_bam = "samtools view -Sbh hits.sam >| {output}"
if not is_there("hits.sam"):
do.run(cmd.format(**locals()), "")
if not is_there(output):
do.run(cmd_bam.format(**locals()), "")
mirbase_output = _annotate(output, mirbase)
return (mirbase_output, full(output))
def _bowtie2(input, index, mirbase):
safe_makedir("bowtie2")
with ch_directory("bowtie2"):
output = "bowtie2_map.bam"
cmd = ("bowtie2 -f -k 50 -L 18 -x {index}"
" -U {input}"
" >| hits.sam")
cmd_bam = "samtools view -Sbh hits.sam >| {output}"
if not is_there("hits.sam"):
do.run(cmd.format(**locals()), "")
if not is_there(output):
do.run(cmd_bam.format(**locals()), "")
mirbase_output = _annotate(output, mirbase)
return (mirbase_output, full(output))
def _annotate(input, mirbase):
output = "mirbase.bed"
cmd = ("bedtools intersect -bed -wo -s -f 0.80 -a"
" {input} -b {mirbase} >| {output}")
if not is_there(output):
do.run(cmd.format(**locals()), "")
return full(output)
def _annotate(input, mirbase):
output = "mirbase.bed"
cmd = ("bedtools intersect -bed -wo -s -f 0.80 -a"
" {input} -b {mirbase} >| {output}")
if not is_there(output):
do.run(cmd.format(**locals()), "")
return full(output)
def _star(input, index, mirbase):
with ch_directory("star"):
output = "star_map.bam"
cmd = ("STAR --genomeDir {index} --readFilesIN {input}"
" --outFilterMultimapNmax 50"
" --outSAMattributes NH HI NM"
" --alignIntronMax 1")
cmd_bam = "samtools view -Sbh Aligned.out.sam >| {output}"
if not is_there("Aligned.out.sam"):
do.run(cmd.format(**locals()), "")
if not is_there(output):
do.run(cmd_bam.format(**locals()), "")
mirbase_output = _annotate(output, mirbase)
return mirbase_output
def _stats(ann, fasta, prefix):
output = prefix + ".tsv"
if not is_there(output):
sim_data = res.read_sim_fa(fasta)
data, counts = res.read_bam(ann)
res.print_output(data, counts, sim_data, output, prefix)
return output
def _star(input, index, mirbase):
safe_makedir("star")
with ch_directory("star"):
output = "star_map.bam"
cmd = ("STAR --genomeDir {index} --readFilesIN {input}"
" --outFilterMultimapNmax 50"
" --outSAMattributes NH HI NM"
" --alignIntronMax 1")
cmd_bam = "samtools view -Sbh Aligned.out.sam >| {output}"
if not is_there("Aligned.out.sam"):
do.run(cmd.format(**locals()), "")
if not is_there(output):
do.run(cmd_bam.format(**locals()), "")
mirbase_output = _annotate(output, mirbase)
return (mirbase_output, full(output))
def _stats(ann, bam, fasta, prefix):
output = prefix + ".tsv"
if not is_there(output):
sim_data = res.read_sim_fa(fasta)
data, counts = res.read_ann(ann)
mapped = res.read_bam(bam)
res.print_output(data, counts, sim_data, mapped, output, prefix)
return output
def _srnamapper(input, index, mirbase):
safe_makedir("srnamapper")
with ch_directory("srnamapper"):
output = "srnamapper_map.bed"
cmd = ("sRNAmapper.pl -i {input} -g {index}"
" -s 10 -n 3 -o hits.eland")
if not is_there("hits.eland"):
do.run(cmd.format(**locals()), "")
if not is_there(output):
with open(output, 'w') as out_handle:
with open("hits.eland") as in_handle:
for line in in_handle:
cols = line.strip().split('\t')
print >>out_handle, "%s\t%s\t%s\t%s\t1\t%s\t1\t1\t1\t1\t1\t1" % (cols[0], cols[1], int(cols[1]) + len(cols[2]), cols[3], cols[6])
mirbase_output = _annotate(output, mirbase)
return (mirbase_output, full(output))
def _srnamapper(input, index, mirbase):
safe_makedir("srnamapper")
with ch_directory("srnamapper"):
output = "srnamapper_map.bed"
cmd = ("sRNAmapper.pl -i {input} -g {index}"
" -s 10 -n 3 -o hits.eland")
if not is_there("hits.eland"):
do.run(cmd.format(**locals()), "")
if not is_there(output):
with open(output, 'w') as out_handle:
with open("hits.eland") as in_handle:
for line in in_handle:
cols = line.strip().split('\t')
print >> out_handle, "%s\t%s\t%s\t%s\t1\t%s\t1\t1\t1\t1\t1\t1" % (
cols[0], cols[1], int(cols[1]) + len(cols[2]),
cols[3], cols[6])
mirbase_output = _annotate(output, mirbase)
return (mirbase_output, full(output))