Strip AS4 annotations from a genbank file

strip_as4_annotations.py

#!/usr/bin/env python

from __future__ import print_function, division

import sys

from helperlibs.bio import seqio

def strip_record(seq_record):
    new_features = []

    for feature in seq_record.features:
        # Discard features added by antiSMASH
        if feature.type in ('cluster', 'cluster_border', 'CDS_motif', 'aSDomain', 'PFAM_domain', 'promoter'):
            continue

        # clean up antiSMASH annotations in CDS features
        if feature.type == 'CDS':
            if 'sec_met' in feature.qualifiers:
                del feature.qualifiers['sec_met']
            if "note" in feature.qualifiers:
                feature.qualifiers["note"] = [note for note in feature.qualifiers["note"] if not note.startswith("smCOG")]
                if not feature.qualifiers["note"]:
                    del feature.qualifiers["note"]

        new_features.append(feature)

    seq_record.features = new_features


def strip_all(filename):
    record = seqio.read(filename)
    strip_record(record)
    seqio.write(record, filename, 'genbank')

if __name__ == "__main__":
    strip_all(sys.argv[1])