make fastq; align; make expression table;

3reads.py

# usage:
#   python 3reads.py /batch/dir1/ /batch/dir2/        # run all steps on two batches
#   python 3reads.py -2 -3 -loadAndKeep /batch/dir    # run steps 2 and 3 and use STAR --loadAndKeep
#   python 3reads.py /batch/dir1 /batch/dir2/ -13     # run steps 1 and 3 on two batches
import sys, os

base_dir = '/home/sxv/code/s7s/'
MAKE_FASTQ = '%s/make_fastq.sh' % base_dir
ALIGN = '%s/align_smart-3seq.sh' % base_dir
MAKE_EXPRESSION_TABLE = '%s/make_expression_table.R' % base_dir

steps = '123'
batch_dirs = []

for arg in sys.argv[1:]:
  if arg[0] == '-':
    if arg == '-loadAndKeep' or arg == '-keep':
      ALIGN = '%s/align_smart-3seq.loadAndKeep.sh' % base_dir
    else:
      if steps == '123':
        steps = arg[1:]
      else:
        steps += arg[1:]
  else:
    batch_dirs.append(arg)

def make_fastq():
  os.system("bash -c 'cp %s %s'" % (MAKE_FASTQ, batch_dir))
  os.system("bash -c 'cd %s && bash %s .'" % (batch_dir, MAKE_FASTQ))
  os.system("bash -c 'cd %s && mkdir -p fastq && mv *gz fastq'" % batch_dir)

def align():
  os.system("bash -c 'cp %s %s/fastq/'" % (ALIGN, batch_dir))
  os.system("bash -c 'cd %s/fastq && bash %s /media/stroma-common/genome/hg38/star/dbsnp147_gencode25-68/ *fastq.gz'" % (batch_dir, ALIGN))
  os.system("bash -c 'cd %s && mkdir -p bam && mv fastq/*ba? fastq/*log bam/'" % batch_dir)

def make_expression_table():
  os.system("bash -c 'cp %s %s/bam'" % (MAKE_EXPRESSION_TABLE, batch_dir))
  os.system("bash -c 'cd %s/bam && Rscript %s --no-rlog /media/stroma-common/genome/hg38/gencode.v25.annotation.gtf *bam'" % (batch_dir, MAKE_EXPRESSION_TABLE))
  os.system("bash -c 'cd %s/bam && xlsx2csv -s 1 gene_expression.xlsx raw_reads.csv'" % batch_dir)
  project_name = batch_dir.split('/')[-1] or batch_dir.split('/')[-2]
  os.system("bash -c 'cd %s/bam && cp raw_reads.csv %s.raw_reads.csv && cp raw_reads.csv %sresults/%s.raw_reads.csv'" % (batch_dir, project_name, base_dir, project_name))

for batch_dir in batch_dirs:
  if '1' in steps: make_fastq()
  if '2' in steps: align()
  if '3' in steps: make_expression_table()
  # if 'o' in steps: optional_step()