def load_contigs(cls, handle, contigswrapper):
"""
Attempts to load contigs to Blazegraph.
Args:
contigswrapper: a ContigsWrapper instance for storing contig
metadata
"""
sequencedata = SeqIO.parse(handle, 'fasta')
contigs = [] # list of 2-tuples (accession name, seq)
for record in sequencedata:
accession_name = record.name.split("|")[3].split(".")[0]
if "complete" in record.description.lower():
accession_name = accession_name + "_closed"
if check_named_individual(accession_name):
print "%s already in Blazegraph." % accession_name
raise TypeError
else:
contigs.append((accession_name, str(record.seq)))
if "plasmid" in record.description.lower():
contigswrapper.dict["is_from"] = 'PLASMID'
contigswrapper.add_contigs(contigs)
def load_sequence(self, seqdata):
"""Checks to see if the sequence is already uploaded onto Blazegraph,
and if not, try loading the sequence.
Logs all TypeErrors (assumption: all other errors are mistakes in the
process and not something that requires
manual curation).
Args:
seqdata: a SequenceMetadata instance storing sequence-related data
that would otherwise be a data clump
"""
print seqdata.name
if check_named_individual(seqdata.name):
print "%s already in Blazegraph." % seqdata.name
raise TypeError
else:
self.get_seqdata(seqdata)
def create_gene(cls, metadata):
"""
Creates a Gene object to export the data out in the turtle format
with the appropriate RDF tags and
uploads it into Blazegraph.
Args:
metadata(GeneMetadata): An instance that contains metadata
pertaining to an individual gene
"""
graph = Graph()
name = metadata.name
if check_named_individual(name):
print "%s already in Blazegraph." % name
else:
kwargs = metadata.build_kwargs()
Gene(graph, **kwargs).rdf()
BlazegraphUploader().upload_data(generate_output(graph))
