def from_gff(cls, filename, name, description, build, organism):
''' imports refgen from gff file '''
self = cls.create(name, description, type='RefGen')
self._global('build', build)
self._global('organism', organism)
genes = list()
with open(filename, 'r') as IN:
for line in IN:
#skip comment lines
if line.startswith('#'):
continue
chrom, source, feature, start, end, score, strand, frame, attributes = line.strip(
).split()
attributes = dict([
(field.split('='))
for field in attributes.strip(';').split(';')
])
if feature == 'chromosome':
self.log('Found a chromosome: {}', attributes['ID'])
self.add_chromosome(Chrom(attributes['ID'], end))
if feature == 'gene':
genes.append(
Gene(chrom,
int(start),
int(end),
attributes['ID'].upper(),
strand=strand,
build=build,
organism=organism))
self.add_gene(genes)
return self
def from_ids(self, gene_list, check_shape=False, enumerated=False):
''' returns gene object list from an iterable of id strings '''
genes = [
Gene(*x, build=self.build, organism=self.organism)
for x in self.db.cursor().execute('''
SELECT chromosome,start,end,id FROM genes WHERE id IN ('{}')
'''.format("','".join(map(str.upper, gene_list))))
]
if check_shape and len(genes) != len(gene_list):
raise ValueError('Some input ids do not have genes in reference')
return genes
def build_cob(args):
try:
# Build the refgen
refgen = co.RefGen(args.refgen)
# Check that the sep is likely right.
if len(pd.read_table(args.filename, sep=args.sep).columns) == 1:
print(("Detected only 1 column in {}, are you sure "
"colunms are separated by '{}'?").format(
args.filename, args.sep))
return None
if args.allow_non_membership:
refgen = refgen.copy('{}_tmp'.format(refgen.name),
'temp refgen'.format(refgen.name))
# Add non membership genes
for gid in pd.read_table(args.filename, sep=args.sep).index:
refgen.add_gene(Gene(None, None, id=gid))
quality_control = False if args.skip_quality_control else True
normalize = False if args.skip_normalization else True
quantile = False if args.skip_quantile else True
# Check to see if this dataset is already built
if available_datasets('Expr', args.name):
print('Warning! This dataset has already been built.')
co.Tools.del_dataset('Expr', args.name, force=args.force)
# Basically just pass all the CLI arguments to the COB class method
cob = co.COB.from_table(
args.filename,
args.name,
args.description,
refgen,
# Optional arguments
sep=args.sep,
rawtype=args.rawtype,
# Data Processing
quality_control=quality_control,
normalization=normalize,
quantile=quantile,
# Data processing parameters
max_gene_missing_data=args.max_gene_missing_data,
max_accession_missing_data=args.max_accession_missing_data,
min_single_sample_expr=args.min_single_sample_expr,
min_expr=args.min_expr,
max_val=args.max_val,
dry_run=args.dry_run,
zscore_cutoff=args.zscore_cutoff,
index_col=args.index_col)
print(cob.summary())
except Exception as e:
print("Build failed. Rolling back: removing corrupted files...")
co.Tools.del_dataset('Expr', args.name, force=True)
raise e
def __getitem__(self, item):
try:
gene_data = self.db.cursor().execute(
'''
SELECT chromosome,start,end,id FROM genes WHERE id = ?
''', (item, )).fetchone()
return Gene(*gene_data, build=self.build, organism=self.organism)
except Exception as e:
pass
try:
_ = (x for x in item)
return list(self.from_ids(list(_)))
except TypeError as e:
self.log('not iterable: {}', e)
pass
return None
def downstream_genes(self, locus, gene_limit=1000):
'''
returns genes downstream of a locus. Genes are ordered so that the
nearest genes are at the beginning of the list.
'''
return [
Gene(*x, build=self.build, organism=self.organism)
for x in self.db.cursor().execute(
'''
SELECT chromosome,start,end,id FROM genes
WHERE chromosome = ?
AND start > ?
AND start < ?
ORDER BY start ASC
LIMIT ?
''', (locus.chrom, locus.start, locus.downstream, gene_limit))
]
def within_gene(self, locus):
'''
Returns the gene the locus is within, or None
'''
try:
x = [
Gene(*x, build=self.build, organism=self.organism)
for x in self.db.cursor().execute(
'''
SELECT chromosome,start,end,id FROM genes
WHERE chromosome = ?
AND start < ?
AND end > ?
''', (locus.chrom, locus.start, locus.start))
][0]
return x
except Exception as e:
return None
def genes_within(self, loci, chain=True):
'''
Returns the genes within a locus, or None
'''
try:
iterator = iter(loci)
genes = [
self.genes_within(locus, chain=chain) for locus in iterator
]
if chain:
genes = list(itertools.chain(*genes))
return genes
except TypeError as e:
return [
Gene(*x, build=self.build, organism=self.organism)
for x in self.db.cursor().execute(
'''
SELECT chromosome,start,end,id FROM genes
WHERE chromosome = ?
AND start > ?
AND end < ?
''', (loci.chrom, loci.start, loci.end))
]
def build_cob(args):
try:
# Build the refgen
refgen = co.RefGen(args.refgen)
# Check that the sep is likely right.
if len(pd.read_table(args.filename, sep=args.sep).columns) == 1:
print(("Detected only 1 column in {}, are you sure "
"colunms are separated by '{}'?").format(
args.filename, args.sep))
return None
elif (len(pd.read_table(args.filename, sep=args.sep).columns) < 20
and args.non_interactive != True):
print((
"Detected fewer than 20 accessions in the expression matrix. "
"Calculating co-expression with this many datapoints is not advised"
))
if input("are you sure you want to continue? [y/n]: ").upper(
) == "Y":
pass
else:
sys.exit(1)
if args.allow_non_membership:
refgen = refgen.copy("{}_tmp".format(refgen.name),
"temp refgen".format(refgen.name))
# Add non membership genes
for gid in pd.read_table(args.filename, sep=args.sep).index:
refgen.add_gene(Gene(None, None, id=gid))
quality_control = False if args.skip_quality_control else True
normalize = False if args.skip_normalization else True
quantile = True if args.quantile else False
# Check to see if this dataset is already built
if available_datasets("Expr", args.name):
print("Warning! This dataset has already been built.")
co.Tools.del_dataset("Expr", args.name, force=args.force)
# Basically just pass all the CLI arguments to the COB class method
cob = co.COB.from_table(
args.filename,
args.name,
args.description,
refgen,
# Optional arguments
sep=args.sep,
rawtype=args.rawtype,
# Data Processing
quality_control=quality_control,
normalization=normalize,
quantile=quantile,
# Data processing parameters
max_gene_missing_data=args.max_gene_missing_data,
max_accession_missing_data=args.max_accession_missing_data,
min_single_sample_expr=args.min_single_sample_expr,
min_expr=args.min_expr,
max_val=args.max_val,
dry_run=args.dry_run,
zscore_cutoff=args.zscore_cutoff,
index_col=args.index_col,
)
print(cob.summary())
except Exception as e:
print(
f"Build failed for {args.name}. Rolling back: removing corrupted files..."
)
co.Tools.del_dataset("Expr", args.name, force=True)
raise e
def iter_genes(self):
''' iterates over genes in refgen, only returns genes within gene filter '''
return (Gene(*x, build=self.build, organism=self.organism)
for x in self.db.cursor().execute('''
SELECT chromosome,start,end,id FROM genes
'''))
def random_gene(self):
return Gene(*self.db.cursor().execute(
'''
SELECT chromosome,start,end,id from genes WHERE rowid = ?
''', (random.randint(1, self.num_genes()), )).fetchone())