def run_stringtie_expression(data):
"""
estimate expression from Stringtie, using the bcbio datadict
does not do transcriptome assembly
"""
bam = dd.get_work_bam(data)
sample_name = dd.get_sample_name(data)
out_dir = os.path.join("stringtie", sample_name)
isoform_fpkm = os.path.join(out_dir, sample_name + ".isoform.fpkm")
gene_fpkm = os.path.join(out_dir, sample_name + ".fpkm")
assembly = os.path.abspath(os.path.join(out_dir, "stringtie-assembly.gtf"))
if file_exists(isoform_fpkm) and file_exists(gene_fpkm):
data = dd.set_stringtie_dir(data, out_dir)
data = dd.set_fpkm(data, gene_fpkm)
data = dd.set_fpkm_isoform(data, isoform_fpkm)
if "stringtie" in dd.get_transcript_assembler(data):
assembled_gtfs = dd.get_assembled_gtf(data)
assembled_gtfs.append(assembly)
data = dd.set_assembled_gtf(data, assembled_gtfs)
return data
with file_transaction(data, out_dir) as tx_out_dir:
transcript_file = _stringtie_expression(bam, data, tx_out_dir)
df = _parse_ballgown(transcript_file)
_write_fpkms(df, tx_out_dir, sample_name)
data = dd.set_stringtie_dir(data, out_dir)
data = dd.set_fpkm(data, gene_fpkm)
data = dd.set_fpkm_isoform(data, isoform_fpkm)
if "stringtie" in dd.get_transcript_assembler(data):
assembled_gtfs = dd.get_assembled_gtf(data)
assembled_gtfs.append(assembly)
data = dd.set_assembled_gtf(data, assembled_gtfs)
return data
def run_stringtie_expression(data):
"""
estimate expression from Stringtie, using the bcbio datadict
does not do transcriptome assembly
"""
bam = dd.get_work_bam(data)
sample_name = dd.get_sample_name(data)
out_dir = os.path.join("stringtie", sample_name)
isoform_fpkm = os.path.join(out_dir, sample_name + ".isoform.fpkm")
gene_fpkm = os.path.join(out_dir, sample_name + ".fpkm")
assembly = os.path.abspath(os.path.join(out_dir, "stringtie-assembly.gtf"))
if file_exists(isoform_fpkm) and file_exists(gene_fpkm):
data = dd.set_stringtie_dir(data, out_dir)
data = dd.set_fpkm(data, gene_fpkm)
data = dd.set_fpkm_isoform(data, isoform_fpkm)
if "stringtie" in dd.get_transcript_assembler(data):
assembled_gtfs = dd.get_assembled_gtf(data)
assembled_gtfs.append(assembly)
data = dd.set_assembled_gtf(data, assembled_gtfs)
return data
with file_transaction(data, out_dir) as tx_out_dir:
transcript_file = _stringtie_expression(bam, data, tx_out_dir)
df = _parse_ballgown(transcript_file)
_write_fpkms(df, tx_out_dir, sample_name)
data = dd.set_stringtie_dir(data, out_dir)
data = dd.set_fpkm(data, gene_fpkm)
data = dd.set_fpkm_isoform(data, isoform_fpkm)
if "stringtie" in dd.get_transcript_assembler(data):
assembled_gtfs = dd.get_assembled_gtf(data)
assembled_gtfs.append(assembly)
data = dd.set_assembled_gtf(data, assembled_gtfs)
return data
def cufflinks_merge(*samples):
to_merge = filter_missing([dd.get_assembled_gtf(data) for data in
dd.sample_data_iterator(samples)])
data = samples[0][0]
bam_file = dd.get_work_bam(data)
ref_file = dd.get_sam_ref(data)
gtf_file = dd.get_gtf_file(data)
out_dir = os.path.join(dd.get_work_dir(data), "assembly")
num_cores = dd.get_num_cores(data)
merged_gtf = cufflinks.merge(to_merge, ref_file, gtf_file, num_cores, samples[0][0])
for data in dd.sample_data_iterator(samples):
dd.set_assembled_gtf(data, merged_gtf)
return samples
def cufflinks_assemble(data):
bam_file = dd.get_work_bam(data)
ref_file = dd.get_sam_ref(data)
out_dir = os.path.join(dd.get_work_dir(data), "assembly")
num_cores = dd.get_num_cores(data)
assembled_gtf = cufflinks.assemble(bam_file, ref_file, num_cores, out_dir, data)
data = dd.set_assembled_gtf(data, assembled_gtf)
return [[data]]
def cufflinks_assemble(data):
bam_file = dd.get_work_bam(data)
ref_file = dd.get_sam_ref(data)
out_dir = os.path.join(dd.get_work_dir(data), "assembly")
num_cores = dd.get_num_cores(data)
assembled_gtf = cufflinks.assemble(bam_file, ref_file, num_cores, out_dir, data)
data = dd.set_assembled_gtf(data, assembled_gtf)
return [[data]]
def cufflinks_merge(*samples):
to_merge = filter_missing([dd.get_assembled_gtf(data) for data in
dd.sample_data_iterator(samples)])
data = samples[0][0]
bam_file = dd.get_work_bam(data)
ref_file = dd.get_sam_ref(data)
gtf_file = dd.get_gtf_file(data)
out_dir = os.path.join(dd.get_work_dir(data), "assembly")
num_cores = dd.get_num_cores(data)
merged_gtf = cufflinks.merge(to_merge, ref_file, gtf_file, num_cores, samples[0][0])
updated_samples = []
for data in dd.sample_data_iterator(samples):
data = dd.set_assembled_gtf(data, merged_gtf)
updated_samples.append([data])
return updated_samples
