def print_sample_statistics(args):
print("Counting statistics")
sample_name = args.sampleName
mate_handler = MateHandler()
sample = mate_handler.get_sample_by_name(sample_name)
total_first = 0
total_second = 0
total_aligned = 0
total_deduplicated = 0
for mate in sample.mates:
di = MateDirInfo(mate)
total_first += count_fastq_reads(di.first_fastqc_zip_path)
total_second += count_fastq_reads(di.second_fastqc_zip_path)
alignment_report = BismarkReport.fromfile(di.alignment_report_path)
total_aligned += alignment_report.num_of_aligned
drp = DeduplicationReportParser(di.deduplication_report_path)
total_deduplicated += drp.sequences_left()
print("Total first: ", total_first)
print("Total second: ", total_second)
print("Total algined: ", total_aligned)
print("Total deduplicated: ", total_deduplicated)
def merge_sample_alignments(args):
sample_name = args.sampleName
mate_handler = MateHandler()
sample = mate_handler.get_sample_by_name(sample_name)
sp = SamplePipeline(sample)
pipeline_handler = PipelineHandler(sp, ncores=22, memory=220000, clean_output_dir=True)
pipeline_handler.run_on_condor()
def align_sample(args):
sample_name = args.sampleName
mate_handler = MateHandler()
sample = mate_handler.get_sample_by_name(sample_name)
## conditionally, submit all the samples directly to condor (remove in the future)
for mate in sample.mates:
pipeline = MatePipeline(mate)
pipeline.setup()
pipeline_handler = PipelineHandler(pipeline, ncores=10, memory=10900)
pipeline_handler.run_on_condor()
print("*************************************", file=self.log_file_handler)
print("Splitting:", file=self.log_file_handler)
self.split_bam_by_chromosome()
print("*************************************", file=self.log_file_handler)
print("", file=self.log_file_handler)
print("", file=self.log_file_handler)
print("*************************************", file=self.log_file_handler)
print("Extracting:", file=self.log_file_handler)
self.extract_methylation()
print("*************************************", file=self.log_file_handler)
print("", file=self.log_file_handler)
print("", file=self.log_file_handler)
# destroy the logger
self.log_file_handler.close()
if __name__ == "__main__":
print("Running sample pipeline for test sample")
mate_handler = MateHandler("Config/samples_config.yaml")
sample = mate_handler.get_sample_by_name("test_sample")
dir_handler = SampleDirInfo(sample)
# print(bamtools_clean_command(dir_handler.aligned_bam_path, dir_handler.filtered_bam_path))
# dir_handler = SampleDirInfo(sample)
# print(dir_handler.list_aligned_bam_files())
sp = SamplePipeline(sample)
sp.pipeline()
class FastqcReportWriter(object):
def __init__(self, fastqc_summary, output_dir="./Fastqc_Writer_Report"):
self.fastqc_summary = fastqc_summary
self.file_manager = FastqcReportFileManager(output_dir)
def write_report_for_tag(self, tag):
file_path = self.file_manager.file_path_for_tag(tag)
html_string = self.fastqc_summary.get_html_for_tag(tag)
with open(file_path, "w") as html_file:
print(html_string, file=html_file)
print("Written: " + os.path.basename(file_path))
def write(self):
for tag in fastqc_report_tags():
self.write_report_for_tag(tag)
if __name__ == "__main__":
mateHandler = MateHandler()
dirInfos = [MateDirInfo(mate) for mate in mateHandler.list_project_mates()]
pairs = list(zip([dirInfo.first_fastqc_zip_path for dirInfo in dirInfos],
[dirInfo.second_fastqc_zip_path for dirInfo in dirInfos]))
fastqcSummary = PairedFastqcSummary(pairs)
writer = FastqcReportWriter(fastqcSummary, "TestData/FastqcSummary")
writer.write()
call_fastqc(self.dir_info.second, self.dir_info.fastqc_output_dir)
except:
raise Exception("Couldn't deduplicate for mate " + self.name() + "! Please inspect the log file")
def call_trim_galore(self):
try:
trimmer = TrimGalore(self.dir_info.first,
self.dir_info.second,
self.dir_info.trim_galore_output_dir)
run_shell_command(trimmer.generate_command())
except:
raise Exception("Couldn't trim galore for mate " + self.name() + "! Please inspect the log file")
def pipeline(self, clean_output_dir=True):
self.setup(clean_output_dir)
self.call_fastqc()
self.call_trim_galore()
self.call_bismark()
self.call_deduplicate()
##self.extract_methylation()
def clean(self):
raise NotImplementedError()
if __name__ == "__main__":
handler = MateHandler()
test_mate = handler.get_mate_by_name("S112_tag_6_GCCAAT_L003_14101")
pipeline = MatePipeline(test_mate)
pipeline.pipeline()
