def main():
f = omim.input(sys.argv[1])
av_count = 0
for record in f:
if record.allelic_variants is not None:
av_count += len(record.allelic_variants)
print av_count
def main():
# return if we don't have the correct arguments
if len(sys.argv) < 2:
raise SystemExit(__doc__.replace("%prog", sys.argv[0]))
f = omim.input(sys.argv[1])
av_count = del_count = dup_count = ins_count = ivs_count = fs_count = unknown_count = 0
for record in f:
if record.allelic_variants is not None:
av_count += len(record.allelic_variants)
for variant in record.allelic_variants:
if variant.mutation is not None:
if variant.mutation.find("DEL") >= 0:
del_count += 1
elif variant.mutation.find("DUP") >= 0:
dup_count += 1
elif variant.mutation.find("INS") >= 0:
ins_count += 1
elif variant.mutation.find("IVS") >= 0:
ivs_count += 1
elif variant.mutation.find("FS") >= 0:
fs_count += 1
else:
# if we don't have a specified point mutation, class it as unknown
try:
aa_pos = int(variant.mutation[3:-3])
except ValueError:
unknown_count += 1
continue
# reformat the reference and mutant amino acids
ref_aa = variant.mutation[0:3].title()
if ref_aa == "Ter":
ref_aa = "TERM"
elif ref_aa not in one_letter_alphabet.keys():
unknown_count += 1
continue
mut_aa = variant.mutation[-3:].title()
if mut_aa == "Ter":
mut_aa = "TERM"
elif mut_aa not in one_letter_alphabet.keys():
unknown_count += 1
continue
# process and print the phenotypes
phenotypes = _process_variant_title(variant.title)
for p in phenotypes:
print "%s\t%s\t%s-%s\t%s\t%s\tomim:%s%s" % (p, variant.gene,
ref_aa, mut_aa, aa_pos, variant.text.count(" ") + 1,
record.number, variant.number)
print >> sys.stderr, "TOTAL VARIANTS PARSED:", av_count, "DEL:", del_count,
print >> sys.stderr, "DUP:", dup_count, "INS:", ins_count,
print >> sys.stderr, "IVS:", ivs_count, "FS:", fs_count,
print >> sys.stderr, "UNKNOWN:", unknown_count
