def test_identity(self):
i1, i2, i3, i4 = P.closed(0, 1), P.openclosed(0, 1), P.closedopen(0, 1), P.open(0, 1)
assert P.from_string(P.to_string(i1), int) == i1
assert P.from_string(P.to_string(i2), int) == i2
assert P.from_string(P.to_string(i3), int) == i3
assert P.from_string(P.to_string(i4), int) == i4
def test_parameters(self):
i1, i2, i3, i4 = P.closed(0, 1), P.openclosed(0, 1), P.closedopen(0, 1), P.open(0, 1)
params = {
'disj': ' or ',
'sep': '-',
'left_open': '<!',
'left_closed': '<',
'right_open': '!>',
'right_closed': '>',
'conv': lambda s: '"{}"'.format(s),
'pinf': '+oo',
'ninf': '-oo',
}
assert P.to_string(i1, **params) == '<"0"-"1">'
assert P.to_string(i2, **params) == '<!"0"-"1">'
assert P.to_string(i3, **params) == '<"0"-"1"!>'
assert P.to_string(i4, **params) == '<!"0"-"1"!>'
assert P.to_string(P.empty(), **params) == '<!!>'
assert P.to_string(P.singleton(1), **params) == '<"1">'
assert P.to_string(P.openclosed(-P.inf, 1), **params) == '<!-oo-"1">'
assert P.to_string(P.closedopen(1, P.inf), **params) == '<"1"-+oo!>'
assert P.to_string(P.closed(0, 1) | P.closed(2, 3), **params) == '<"0"-"1"> or <"2"-"3">'
gtffile = args.gtf
dbfile = args.db
jsonfile = args.json
threads = int(args.threads)
print('Creating gtf database, this will take some time...')
db = gffutils.create_db(gtffile, dbfile)
isoform_interval_dict = {}
for gene in db.features_of_type('gene'):
g_id = gene['gene_id'][0]
isoform_interval_dict[g_id] = {}
for transcript in db.children(gene, featuretype='transcript'):
t_id = transcript['transcript_id'][0]
isoform_interval_dict[g_id][t_id] = P.empty()
for exon in db.children(transcript, featuretype='exon'):
isoform_interval_dict[g_id][
t_id] = isoform_interval_dict[g_id][t_id] | P.closed(
exon.start, exon.end)
print('Extracting unque isoform intervals')
res = Parallel(n_jobs=threads, verbose=3, backend='loky')(
delayed(create_interval_dict_linear_time)(gene, transcript_intervals)
for gene, transcript_intervals in isoform_interval_dict.items())
isoform_unique_intervals = {k: v for k, v in res}
isoform_unique_intervals_for_json_dump = {
gene: {P.to_string(k): ','.join(v)
for k, v in d.items()}
for gene, d in isoform_unique_intervals.items()
}
print('Writing unique isoform intervals to json file {}'.format(jsonfile))
with open(jsonfile, 'w') as fp:
json.dump(isoform_unique_intervals_for_json_dump, fp)
def test_bounds(self):
assert P.to_string(P.closed(0, 1)) == '[0,1]'
assert P.to_string(P.openclosed(0, 1)) == '(0,1]'
assert P.to_string(P.closedopen(0, 1)) == '[0,1)'
assert P.to_string(P.open(0, 1)) == '(0,1)'
def test_bound_types(self):
assert P.to_string(P.closed('a', 'b')) == "['a','b']"
assert P.to_string(P.closed(tuple([0]), tuple([1]))) == '[(0,),(1,)]'
def test_unions(self):
assert P.to_string(P.closed(0, 1) | P.closed(2, 3)) == '[0,1] | [2,3]'
def test_infinities(self):
assert P.to_string(P.openclosed(-P.inf, 1)) == '(-inf,1]'
assert P.to_string(P.closedopen(1, P.inf)) == '[1,+inf)'
def test_empty(self):
assert P.to_string(P.empty()) == '()'
def test_singleton(self):
assert P.to_string(P.singleton(0)) == '[0]'
def time_interval_to_str(interval):
converter = lambda t: t.strftime(time_format)
temp_interval = interval.apply(lambda i:
(P.CLOSED, i.lower, i.upper, P.CLOSED))
return P.to_string(temp_interval, conv=converter)
