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()
