def wiggle(self):
jobs = []
for sample in self.samples:
bam_file_prefix = os.path.join("alignment", sample.name, sample.name + ".merged.mdup.")
input_bam = bam_file_prefix + "bam"
bed_graph_prefix = os.path.join("tracks", sample.name, sample.name)
big_wig_prefix = os.path.join("tracks", "bigWig", sample.name)
if config.param('tophat', 'strandInfo') != 'fr-unstranded':
input_bam_f1 = bam_file_prefix + "tmp1.forward.bam"
input_bam_f2 = bam_file_prefix + "tmp2.forward.bam"
input_bam_r1 = bam_file_prefix + "tmp1.reverse.bam"
input_bam_r2 = bam_file_prefix + "tmp2.reverse.bam"
output_bam_f = bam_file_prefix + "forward.bam"
output_bam_r = bam_file_prefix + "reverse.bam"
bam_f_job = concat_jobs([
samtools.view(input_bam, input_bam_f1, "-bh -F 256 -f 81"),
samtools.view(input_bam, input_bam_f2, "-bh -F 256 -f 161"),
picard.merge_sam_files([input_bam_f1, input_bam_f2], output_bam_f),
Job(command="rm " + input_bam_f1 + " " + input_bam_f2)
], name="wiggle." + sample.name + ".forward_strandspec")
bam_r_job = concat_jobs([
Job(command="mkdir -p " + os.path.join("tracks", sample.name) + " " + os.path.join("tracks", "bigWig")),
samtools.view(input_bam, input_bam_r1, "-bh -F 256 -f 97"),
samtools.view(input_bam, input_bam_r2, "-bh -F 256 -f 145"),
picard.merge_sam_files([input_bam_r1, input_bam_r2], output_bam_r),
Job(command="rm " + input_bam_r1 + " " + input_bam_r2)
], name="wiggle." + sample.name + ".reverse_strandspec")
jobs.extend([bam_f_job, bam_r_job])
outputs = [
[bed_graph_prefix + ".forward.bedGraph", big_wig_prefix + ".forward.bw"],
[bed_graph_prefix + ".reverse.bedGraph", big_wig_prefix + ".reverse.bw"],
]
else:
outputs = [[bed_graph_prefix + ".bedGraph", big_wig_prefix + ".bw"]]
for bed_graph_output, big_wig_output in outputs:
job = concat_jobs([
Job(command="mkdir -p " + os.path.join("tracks", sample.name) + " " + os.path.join("tracks", "bigWig")),
bedtools.graph(input_bam, bed_graph_output, big_wig_output)
], name="wiggle." + re.sub(".bedGraph", "", os.path.basename(bed_graph_output)))
jobs.append(job)
return jobs
def picard_merge_sam_files(self):
jobs = []
for sample in self.samples:
# Skip samples with one readset only, since symlink has been created at align step
if len(sample.readsets) > 1:
alignment_directory = os.path.join("alignment", sample.name)
inputs = [os.path.join(alignment_directory, readset.name + ".sorted.bam") for readset in sample.readsets]
output = os.path.join(alignment_directory, sample.name + ".sorted.bam")
job = picard.merge_sam_files(inputs, output)
job.name = "picard_merge_sam_files." + sample.name
jobs.append(job)
return jobs
def merge_realigned(self):
jobs = []
nb_realign_jobs = config.param('indel_realigner', 'nbRealignJobs', type='posint')
for sample in self.samples:
alignment_directory = os.path.join("alignment", sample.name)
realign_directory = os.path.join(alignment_directory, "realign")
merged_realigned_bam = os.path.join(alignment_directory, sample.name + ".realigned.qsorted.bam")
# if nb_realign_jobs == 1, symlink has been created in indel_realigner and merging is not necessary
if nb_realign_jobs > 1:
realigned_bams = [os.path.join(realign_directory, sequence['name'] + ".bam") for sequence in self.sequence_dictionary[0:min(nb_realign_jobs - 1, len(self.sequence_dictionary))]]
realigned_bams.append(os.path.join(realign_directory, "others.bam"))
job = picard.merge_sam_files(realigned_bams, merged_realigned_bam)
job.name = "merge_realigned." + sample.name
jobs.append(job)
return jobs