nickloman · June 9, 2014 16:01 · Jun 9, 2014
diff --git a/fast5tofasta.py b/fast5tofasta.py
@@ -0,0 +1,20 @@
+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()