def cluster(seqfile, seqnames, identity=1.0, prefix='cluster-', threads=None):
with util.ntf(prefix=prefix, suffix='.fasta') as fa, \
util.ntf(prefix=prefix, suffix='.uc') as uc:
wrap.esl_sfetch(seqfile, seqnames, fa)
fa.flush()
uclust.cluster(fa.name,
uc.name,
pct_id=identity,
pre_sorted=False,
quiet=True,
threads=threads)
df = uclust.parse_uclust_as_df(uc)
df = df[df.type != 'C']
df = df[['type', 'query_label', 'target_label']]
return df
def cluster(seqfile, seqnames, identity=1.0, prefix='cluster-', threads=None):
prefix = prefix.replace('/', '\\') # / confuses the filesystem
with util.ntf(prefix=prefix, suffix='.fasta') as fa, \
util.ntf(prefix=prefix, suffix='.uc') as uc:
wrap.esl_sfetch(seqfile, seqnames, fa)
fa.flush()
uclust.cluster(fa.name,
uc.name,
pct_id=identity,
pre_sorted=False,
quiet=True,
threads=threads)
df = uclust.parse_uclust_as_df(uc)
df = df[df.type != 'C']
df = df[['type', 'query_label', 'target_label']]
return df
def test01(self):
df = uclust.parse_uclust_as_df(self.infile)
# target_label always has a value for types S and H
self.assertFalse(any(df[df['type'] != 'C']['target_label'].isnull()))
