fast5tofasta.py

fast5tofasta.py

import h5py
from Bio import SeqIO
from StringIO import StringIO
import sys

keys = {'template' : '/Analyses/Basecall_2D_000/BaseCalled_template/Fastq',
        'complement' : '/Analyses/Basecall_2D_000/BaseCalled_complement/Fastq',
        'twodirections' : '/Analyses/Basecall_2D_000/BaseCalled_2D/Fastq'}

for fn in sys.argv[1:]:
	hdf = h5py.File(fn, 'r')
	for id, key in keys.iteritems():
		try:
			fq = hdf[key][()]
			rec = SeqIO.read(StringIO(fq), "fastq")
			rec.id += "_" + id
			SeqIO.write([rec], sys.stdout, "fasta")
		except Exception, e:
			pass
	hdf.close()