def produce_analysis_piper(ngi_config, project_id):
#create piper_ngi
analysis_dir = os.path.join(ngi_config["analysis"]["base_root"],
ngi_config["analysis"]["sthlm_root"],
ngi_config["analysis"]["top_dir"],
"ANALYSIS", project_id)
data_dir = os.path.join(ngi_config["analysis"]["base_root"],
ngi_config["analysis"]["sthlm_root"],
ngi_config["analysis"]["top_dir"],
"DATA", project_id)
piper_ngi_dir = os.path.join(analysis_dir, "piper_ngi")
fs.create_folder(piper_ngi_dir)
piper_dirs = ["01_raw_alignments","02_preliminary_alignment_qc","03_genotype_concordance",
"04_merged_aligments","05_processed_alignments","06_final_alignment_qc","07_variant_calls","08_misc"]
for piper_dir in piper_dirs:
current_dir = os.path.join(piper_ngi_dir, piper_dir)
fs.create_folder(current_dir)
if piper_dir == "05_processed_alignments":
for sample_id in os.listdir(data_dir):
bam_file = "{}.clean.dedup.bam".format(sample_id)
fs.touch(os.path.join(current_dir, bam_file))
if piper_dir == "07_variant_calls":
for sample_id in os.listdir(data_dir):
vcf_file = "{}.clean.dedup.recal.bam.raw.indel.vcf.gz".format(sample_id)
fs.touch(os.path.join(current_dir, vcf_file))
current_dir = os.path.join(piper_ngi_dir, "sbatch")
fs.create_folder(current_dir)
current_dir = os.path.join(piper_ngi_dir, "setup_xml_files")
fs.create_folder(current_dir)
current_dir = os.path.join(piper_ngi_dir, "logs")
fs.create_folder(current_dir)
create_version_report(current_dir)
def create_FC(incoming_dir, run_name, samplesheet, fastq_1 = None, fastq_2=None ):
# Create something like 160217_ST-E00201_0063_AHJHNYCCXX
path_to_fc = os.path.join(incoming_dir, run_name)
if os.path.exists(path_to_fc):
# This FC exists, skip it
return
fs.create_folder(path_to_fc)
fs.touch(os.path.join(path_to_fc, 'RTAComplete.txt'))
# Create folder Demultiplexing
fs.create_folder(os.path.join(path_to_fc, 'Demultiplexing'))
# Create folder Demultiplexing/Reports
fs.create_folder(os.path.join(path_to_fc, 'Demultiplexing', 'Reports'))
# Create folder Demultiplexing/Stats
fs.create_folder(os.path.join(path_to_fc, 'Demultiplexing', 'Stats'))
# Memorise SampleSheet stats
header = []
for key in samplesheet[0]:
header.append(key)
counter = 1
current_lane = ''
for line in samplesheet:
project_name = line.get('Sample_Project', line.get('Project', ''))
lane = line['Lane']
if current_lane == '':
current_lane = lane
elif current_lane != lane:
counter = 1
current_lane = lane
sample_id = line.get('SampleID', line.get('Sample_ID', ''))
sample_name = line.get('SampleName', line.get('Sample_Name', ''))
# Create dir structure
fs.create_folder(os.path.join(path_to_fc, 'Demultiplexing', project_name, sample_id))
# Now create the data
fastq_1_dest = '{}_S{}_L00{}_R1_001.fastq.gz'.format(sample_name, counter, lane)
fastq_2_dest = '{}_S{}_L00{}_R2_001.fastq.gz'.format(sample_name, counter, lane)
counter += 1
if fastq_1 is None:
fs.touch(os.path.join(path_to_fc, 'Demultiplexing', project_name,
sample_id, fastq_1_dest))
fs.touch(os.path.join(path_to_fc, 'Demultiplexing', project_name,
sample_id, fastq_2_dest))
else:
fs.do_symlink(fastq_1, os.path.join(path_to_fc, 'Demultiplexing',
project_name, sample_id, fastq_1_dest))
fs.do_symlink(fastq_2, os.path.join(path_to_fc, 'Demultiplexing',
project_name, sample_id, fastq_2_dest))
with open(os.path.join(path_to_fc, 'SampleSheet.csv'), 'w') as Samplesheet_file:
Samplesheet_file.write(u'[Header]\n')
Samplesheet_file.write(u'Date,2016-03-29\n')
Samplesheet_file.write(u'Investigator Name,Christian Natanaelsson\n')
Samplesheet_file.write(u'[Data]\n')
for key in header:
Samplesheet_file.write(u'{},'.format(key))
Samplesheet_file.write(u'\n')
for line in samplesheet:
for key in header:
Samplesheet_file.write(u'{},'.format(line[key]))
Samplesheet_file.write(u'\n')
def produce_analysis_piper(ngi_config, project_id):
# Create piper_ngi
analysis_dir = os.path.join(ngi_config['analysis']['base_root'],
ngi_config['analysis']['sthlm_root'],
ngi_config['analysis']['top_dir'],
'ANALYSIS', project_id)
data_dir = os.path.join(ngi_config['analysis']['base_root'],
ngi_config['analysis']['sthlm_root'],
ngi_config['analysis']['top_dir'],
'DATA', project_id)
piper_ngi_dir = os.path.join(analysis_dir, 'piper_ngi')
fs.create_folder(piper_ngi_dir)
piper_dirs = ['01_raw_alignments',
'02_preliminary_alignment_qc',
'03_genotype_concordance',
'04_merged_aligments',
'05_processed_alignments',
'06_final_alignment_qc',
'07_variant_calls',
'08_misc']
for piper_dir in piper_dirs:
current_dir = os.path.join(piper_ngi_dir, piper_dir)
fs.create_folder(current_dir)
if piper_dir == '05_processed_alignments':
for sample_id in os.listdir(data_dir):
bam_file = '{}.clean.dedup.bam'.format(sample_id)
fs.touch(os.path.join(current_dir, bam_file))
if piper_dir == '07_variant_calls':
for sample_id in os.listdir(data_dir):
vcf_file = '{}.clean.dedup.recal.bam.raw.indel.vcf.gz'.format(sample_id)
fs.touch(os.path.join(current_dir, vcf_file))
current_dir = os.path.join(piper_ngi_dir, 'sbatch')
fs.create_folder(current_dir)
current_dir = os.path.join(piper_ngi_dir, 'setup_xml_files')
fs.create_folder(current_dir)
current_dir = os.path.join(piper_ngi_dir, 'logs')
fs.create_folder(current_dir)
create_version_report(current_dir)
def create_FC(incoming_dir, run_name, samplesheet, fastq_1 = None, fastq_2=None ):
# create something like 160217_ST-E00201_0063_AHJHNYCCXX
path_to_fc = os.path.join(incoming_dir, run_name)
if os.path.exists(path_to_fc):
# this FC exists, skip it
return
fs.create_folder(path_to_fc)
fs.touch(os.path.join(path_to_fc, "RTAComplete.txt"))
# create folder Demultiplexing
fs.create_folder(os.path.join(path_to_fc, "Demultiplexing"))
# create folder Demultiplexing/Reports
fs.create_folder(os.path.join(path_to_fc, "Demultiplexing", "Reports"))
# create folder Demultiplexing/Stats
fs.create_folder(os.path.join(path_to_fc, "Demultiplexing", "Stats"))
#memorise SampleSheet stats
header = []
for key in samplesheet[0]:
header.append(key)
counter = 1
current_lane = ""
for line in samplesheet:
project_name = line.get("Sample_Project", line.get("Project", ""))
lane = line["Lane"]
if current_lane == "":
current_lane = lane
elif current_lane != lane:
counter = 1
current_lane = lane
sample_id = line.get("SampleID", line.get("Sample_ID", ""))
sample_name = line.get("SampleName", line.get("Sample_Name", ""))
#create dir structure
fs.create_folder(os.path.join(path_to_fc, "Demultiplexing", project_name, sample_id))
#now create the data
fastq_1_dest = "{}_S{}_L00{}_R1_001.fastq.gz".format(sample_name, counter, lane)
fastq_2_dest = "{}_S{}_L00{}_R2_001.fastq.gz".format(sample_name, counter, lane)
counter += 1
if fastq_1 is None:
fs.touch(os.path.join(path_to_fc, "Demultiplexing", project_name, sample_id, fastq_1_dest))
fs.touch(os.path.join(path_to_fc, "Demultiplexing", project_name, sample_id, fastq_2_dest))
else:
fs.do_symlink(fastq_1, os.path.join(path_to_fc, "Demultiplexing", project_name, sample_id, fastq_1_dest))
fs.do_symlink(fastq_2, os.path.join(path_to_fc, "Demultiplexing", project_name, sample_id, fastq_2_dest))
with open(os.path.join(path_to_fc, "SampleSheet.csv"), "w") as Samplesheet_file:
Samplesheet_file.write("[Header]\n")
Samplesheet_file.write("Date,2016-03-29\n")
Samplesheet_file.write("Investigator Name,Christian Natanaelsson\n")
Samplesheet_file.write("[Data]\n")
for key in header:
Samplesheet_file.write("{},".format(key))
Samplesheet_file.write("\n")
for line in samplesheet:
for key in header:
Samplesheet_file.write("{},".format(line[key]))
Samplesheet_file.write("\n")
def test_touch(self):
"""Make empty file."""
new_file = os.path.join(self.rootdir, 'empty')
filesystem.touch(new_file)
self.assertTrue(os.path.isfile(new_file))