extract_fasta_local.py
4430ecaacffamaster

extract_fasta_local.py
View Options

			#! /usr/bin/env python3

			import os
			import sys
			import subprocess
			import shutil

			"""
			# get username on SCITAS server
			process = subprocess.run('echo $USER', shell=True,
			stdout=subprocess.PIPE)
			username = process.stdout.decode().rstrip()
			"""

			scratch_path = f"{os.getenv('HOME')}/scratch/"

			temp_path = os.path.join(scratch_path, 'temp')

			split_fasta_path = os.path.join(scratch_path, 'split_fasta')

			if os.path.exists(split_fasta_path):
			shutil.rmtree(split_fasta_path)

			bed_files = os.listdir(temp_path)

			os.mkdir(split_fasta_path)

			"""the reference genome sequence is converted to all upper case (-noMask)
			to avoid any downstream problems with FIMO"""

			for f in bed_files:
			process = subprocess.run(['../ucsc_bin/twoBitToFa',
			'-noMask',
			'-bed={}'.format(os.path.join(temp_path, f)),
			'../data/hg19.2bit',
			os.path.join(split_fasta_path, os.path.splitext(f)[0])+'.fasta'],
			stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
			print(process.stdout)