parse_nctc3000.py

import os
from Bio import SeqIO
from BCBio import GFF

root_dir = '/Users/flashton/projects/nctc3000/2016.01.17'

def main(root_dir):
	for each in os.listdir(root_dir):
		with open('%s/%s' % (root_dir, each)) as fi:
			basename = each.split('.')[0]
			if each.endswith('embl'):
				with open('%s/%s.fa' % (root_dir, basename), 'w') as fo:
					for rec in SeqIO.parse(fi, 'embl'):
						fo.write('>%s\n' % rec.id)
						fo.write('%s\n' % rec.seq)
			
			elif each.endswith('gff'):
				with open('%s/%s.fa' % (root_dir, basename), 'w') as fo:
					try:
						for rec in GFF.parse(fi):
							fo.write('>%s\n' % rec.id)
							fo.write('%s\n' % rec.seq)
					except:
						print each, 'no dice'
		# os.system('rm -rf %s/%s' % (root_dir, each))


main(root_dir)