def get_features(self, ctx, ref, feature_id_list):
"""
Retrieve Feature data.
@param feature_id_list List of Features to retrieve.
If None, returns all Feature data.
@return Mapping from Feature IDs to dicts of available data.
:param ref: instance of type "ObjectReference"
:param feature_id_list: instance of list of String
:returns: instance of mapping from String to type "Feature_data" ->
structure: parameter "feature_id" of String, parameter
"feature_type" of String, parameter "feature_function" of String,
parameter "feature_aliases" of mapping from String to list of
String, parameter "feature_dna_sequence_length" of Long, parameter
"feature_dna_sequence" of String, parameter "feature_md5" of
String, parameter "feature_locations" of list of type "Region" ->
structure: parameter "contig_id" of String, parameter "strand" of
String, parameter "start" of Long, parameter "length" of Long,
parameter "feature_publications" of list of String, parameter
"feature_quality_warnings" of list of String, parameter
"feature_quality_score" of list of String, parameter
"feature_notes" of String, parameter "feature_inference" of String
"""
# ctx is the context object
# return variables are: returnVal
#BEGIN get_features
ga = GenomeAnnotationAPI_local(self.services, ctx['token'], ref)
returnVal = ga.get_features(feature_id_list)
#END get_features
# At some point might do deeper type checking...
if not isinstance(returnVal, dict):
raise ValueError('Method get_features return value ' +
'returnVal is not type dict as required.')
# return the results
return [returnVal]
def get_proteins(self, ctx, ref):
"""
Retrieve Protein data.
@return Mapping from protein ID to data about the protein.
:param ref: instance of type "ObjectReference"
:returns: instance of mapping from String to type "Protein_data" ->
structure: parameter "protein_id" of String, parameter
"protein_amino_acid_sequence" of String, parameter
"protein_function" of String, parameter "protein_aliases" of list
of String, parameter "protein_md5" of String, parameter
"protein_domain_locations" of list of String
"""
# ctx is the context object
# return variables are: returnVal
#BEGIN get_proteins
ga = GenomeAnnotationAPI_local(self.services, ctx['token'], ref)
returnVal = ga.get_proteins()
#END get_proteins
# At some point might do deeper type checking...
if not isinstance(returnVal, dict):
raise ValueError('Method get_proteins return value ' +
'returnVal is not type dict as required.')
# return the results
return [returnVal]
def get_mrna_by_gene(self, ctx, ref, gene_id_list):
"""
Retrieve the mRNA IDs for given gene IDs.
@param gene_id_list List of gene Feature IDS for which to retrieve mRNA IDs.
If empty, returns all gene/mRNA mappings.
@return Mapping of gene Feature IDs to a list of mRNA Feature IDs.
:param ref: instance of type "ObjectReference"
:param gene_id_list: instance of list of String
:returns: instance of mapping from String to list of String
"""
# ctx is the context object
# return variables are: returnVal
#BEGIN get_mrna_by_gene
ga = GenomeAnnotationAPI_local(self.services, ctx['token'], ref)
if not gene_id_list:
returnVal = ga.get_mrna_by_gene([])
else:
returnVal = ga.get_mrna_by_gene(gene_id_list)
#END get_mrna_by_gene
# At some point might do deeper type checking...
if not isinstance(returnVal, dict):
raise ValueError('Method get_mrna_by_gene return value ' +
'returnVal is not type dict as required.')
# return the results
return [returnVal]
def get_feature_type_descriptions(self, ctx, ref, feature_type_list):
"""
Retrieve the descriptions for each Feature type in
this GenomeAnnotation.
@param feature_type_list List of Feature types. If this list
is empty or None,
the whole mapping will be returned.
@return Name and description for each requested Feature Type
:param ref: instance of type "ObjectReference"
:param feature_type_list: instance of list of String
:returns: instance of mapping from String to String
"""
# ctx is the context object
# return variables are: returnVal
#BEGIN get_feature_type_descriptions
ga = GenomeAnnotationAPI_local(self.services, ctx['token'], ref)
returnVal = ga.get_feature_type_descriptions(feature_type_list)
#END get_feature_type_descriptions
# At some point might do deeper type checking...
if not isinstance(returnVal, dict):
raise ValueError('Method get_feature_type_descriptions return value ' +
'returnVal is not type dict as required.')
# return the results
return [returnVal]
def get_mrna_exons(self, ctx, ref, mrna_id_list):
"""
Retrieve Exon information for each mRNA ID.
@param mrna_id_list List of mRNA Feature IDS for which to retrieve exons.
If empty, returns data for all exons.
@return Mapping of mRNA Feature IDs to a list of exons (:js:data:`Exon_data`).
:param ref: instance of type "ObjectReference"
:param mrna_id_list: instance of list of String
:returns: instance of mapping from String to list of type "Exon_data"
-> structure: parameter "exon_location" of type "Region" ->
structure: parameter "contig_id" of String, parameter "strand" of
String, parameter "start" of Long, parameter "length" of Long,
parameter "exon_dna_sequence" of String, parameter "exon_ordinal"
of Long
"""
# ctx is the context object
# return variables are: returnVal
#BEGIN get_mrna_exons
ga = GenomeAnnotationAPI_local(self.services, ctx['token'], ref)
returnVal = ga.get_mrna_by_exons(mrna_id_list)
#END get_mrna_exons
# At some point might do deeper type checking...
if not isinstance(returnVal, dict):
raise ValueError('Method get_mrna_exons return value ' +
'returnVal is not type dict as required.')
# return the results
return [returnVal]
def get_summary(self, ctx, ref):
"""
Retrieve a summary representation of this GenomeAnnotation.
@return summary data
:param ref: instance of type "ObjectReference"
:returns: instance of type "Summary_data" -> structure: parameter
"scientific_name" of String, parameter "taxonomy_id" of Long,
parameter "kingdom" of String, parameter "scientific_lineage" of
list of String, parameter "genetic_code" of Long, parameter
"organism_aliases" of list of String, parameter "assembly_source"
of String, parameter "assembly_source_id" of String, parameter
"assembly_source_date" of String, parameter "gc_content" of
Double, parameter "dna_size" of Long, parameter "num_contigs" of
Long, parameter "contig_ids" of list of String, parameter
"external_source" of String, parameter "external_source_date" of
String, parameter "release" of String, parameter
"original_source_filename" of String, parameter
"feature_type_counts" of mapping from String to Long
"""
# ctx is the context object
# return variables are: returnVal
#BEGIN get_summary
ga = GenomeAnnotationAPI_local(self.services, ctx['token'], ref)
returnVal = ga.get_summary()
#END get_summary
# At some point might do deeper type checking...
if not isinstance(returnVal, dict):
raise ValueError('Method get_summary return value ' +
'returnVal is not type dict as required.')
# return the results
return [returnVal]
def proteins_to_fasta(ws_url='https://ci.kbase.us/services/ws/'):
"""Write FASTA file from a genome reference.
Args:
ws_url: Workspace service URL
Returns:
Full path to output file
"""
ref = "ReferenceGenomeAnnotations/kb|g.166819"
# ref = "ReferenceGenomeAnnotations/kb|g.3899"
genome_annotation = GenomeAnnotationAPI(
token=os.environ.get('KB_AUTH_TOKEN'),
services={'workspace_service_url': ws_url},
ref=ref)
# Get all the proteins with the Data API
proteins = genome_annotation.get_proteins()
# Create an output file and write to it
outfile = '/tmp/166819_prot.fasta'
with open(outfile, 'w') as f:
for fasta_line in get_fasta(proteins):
f.write(fasta_line)
return outfile
def get_cds_by_mrna(self, ctx, ref, mrna_id_list):
"""
Retrieves coding sequence (cds) Feature IDs for given mRNA Feature IDs.
@param mrna_id_list List of mRNA Feature IDS for which to retrieve CDS.
If empty, returns data for all features.
@return Mapping of mRNA Feature IDs to a list of CDS Feature IDs.
:param ref: instance of type "ObjectReference"
:param mrna_id_list: instance of list of String
:returns: instance of mapping from String to String
"""
# ctx is the context object
# return variables are: returnVal
#BEGIN get_cds_by_mrna
ga = GenomeAnnotationAPI_local(self.services, ctx['token'], ref)
if not mrna_id_list:
returnVal = ga.get_cds_by_mrna([])
else:
returnVal = ga.get_cds_by_mrna(mrna_id_list)
#END get_cds_by_mrna
# At some point might do deeper type checking...
if not isinstance(returnVal, dict):
raise ValueError('Method get_cds_by_mrna return value ' +
'returnVal is not type dict as required.')
# return the results
return [returnVal]
def get_feature_locations(self, ctx, ref, feature_id_list):
"""
Retrieve Feature locations.
@param feature_id_list List of Feature IDs for which to retrieve locations.
If empty, returns data for all features.
@return Mapping from Feature IDs to location information for each.
:param ref: instance of type "ObjectReference"
:param feature_id_list: instance of list of String
:returns: instance of mapping from String to list of type "Region" ->
structure: parameter "contig_id" of String, parameter "strand" of
String, parameter "start" of Long, parameter "length" of Long
"""
# ctx is the context object
# return variables are: returnVal
#BEGIN get_feature_locations
ga = GenomeAnnotationAPI_local(self.services, ctx['token'], ref)
returnVal = ga.get_feature_locations(feature_id_list)
#END get_feature_locations
# At some point might do deeper type checking...
if not isinstance(returnVal, dict):
raise ValueError('Method get_feature_locations return value ' +
'returnVal is not type dict as required.')
# return the results
return [returnVal]
def setup():
shared.setup()
global t_new, t_new_e, t_old, t_old_e, t_client_new, t_client_old
t_new = GenomeAnnotationAPI(shared.services, shared.token, genome_new)
t_new_e = _GenomeAnnotation(shared.services, shared.token, genome_new)
t_old = GenomeAnnotationAPI(shared.services, shared.token, genome_old)
t_old_e = _KBaseGenomes_Genome(shared.services, shared.token, genome_old)
t_client_new = GenomeAnnotationClientAPI(
shared.services["genome_annotation_service_url"], shared.token,
genome_new)
t_client_old = GenomeAnnotationClientAPI(
shared.services["genome_annotation_service_url"], shared.token,
genome_old)
def run(ws_url='https://ci.kbase.us/services/ws/'):
#an example KBase reference genome
genomeref = "ReferenceGenomeAnnotations/kb|g.166819"
#creating a new GenomeAnnotation object
genome_annotation = GenomeAnnotationAPI(services = {"workspace_service_url": ws_url}, token=os.environ.get('KB_AUTH_TOKEN'), ref=genomeref)
#return all protein objects from this genome
proteins= genome_annotation.get_proteins()
#format the proteins in FASTA format
fasta = get_protein_fasta(proteins)
outfile = '166819_prot.fasta'
with open(outfile, 'w') as f:
f.write(fasta)
def generate_gff(logger, internal_services, token, ref, output_dir, obj_name,
output_file):
try:
ga = GenomeAnnotationAPI(internal_services,
token=token,
ref="{}/{}".format(ref, obj_name))
except:
raise Exception("Unable to Call GenomeAnnotationAPI : {0}".format(e))
logger.info("Requesting GenomeAnnotation GFF for {}/{}".format(
ref, obj_name))
gff_start = datetime.datetime.utcnow()
#output_file = os.path.join(output_dir,'{}.gff'.format(obj_name))
with open(output_file, 'w') as gff_file:
ga.get_gff().to_file(gff_file)
gff_file.close()
gff_end = datetime.datetime.utcnow()
logger.info("Generating GFF for {} took {}".format(obj_name,
gff_end - gff_start))
def generate_fasta(logger, internal_services, token, ref, output_dir,
obj_name):
try:
ga = GenomeAnnotationAPI(internal_services,
token=token,
ref="{}/{}".format(ref, obj_name))
except Exception as e:
raise Exception(
"Unable to Call GenomeAnnotationAPI : {0}: {1}".format(e))
logger.info("Generating FASTA file from Assembly for {}/{}".format(
ref, obj_name))
fasta_start = datetime.datetime.utcnow()
output_file = os.path.join(output_dir, '{}.fasta'.format(obj_name))
with open(output_file, 'w') as fasta_file:
ga.get_assembly().get_fasta().to_file(fasta_file)
fasta_file.close()
fasta_end = datetime.datetime.utcnow()
logger.info("Generating FASTA for {} took {}".format(
obj_name, fasta_end - fasta_start))
return output_file
def generate_gff(logger,internal_services,token,ref,output_dir,obj_name,output_file):
try:
ga = GenomeAnnotationAPI(internal_services,
token=token,
ref= ref)
except:
raise Exception("Unable to Call GenomeAnnotationAPI : {0}".format(("".join(traceback.format_exc()))))
logger.info("Requesting GenomeAnnotation GFF for {}".format(obj_name))
gff_start = datetime.datetime.utcnow()
gff_file= io.open(output_file, 'wb')
#output_file = os.path.join(output_dir,'{}.gff'.format(obj_name))
try:
ga.get_gff().to_file(gff_file)
except Exception as e:
#raise Exception("Unable to Create GFF file from Genome Annotation : {0}: {1}".format(obj_name,e))
raise Exception("Unable to Create GFF file from Genome Annotation : {0}: {1}".format(obj_name,"".join(traceback.format_exc())))
finally:
gff_file.close()
gff_end = datetime.datetime.utcnow()
logger.info("Generating GFF for {} took {}".format(obj_name, gff_end - gff_start))
def generate_fasta(logger,internal_services,token,ref,output_dir,obj_name):
try:
ga = GenomeAnnotationAPI(internal_services,
token=token,
ref= ref)
except Exception as e:
raise Exception("Unable to Call GenomeAnnotationAPI : {0}".format("".join(traceback.format_exc())))
logger.info("Generating FASTA file from Assembly for {}".format(obj_name))
fasta_start = datetime.datetime.utcnow()
output_file = os.path.join(output_dir,'{}.fasta'.format(obj_name))
fasta_file= io.open(output_file, 'wb')
try:
ga.get_assembly().get_fasta().to_file(fasta_file)
except Exception as e:
#raise Exception("Unable to Create FASTA file from Genome Annotation : {0}".format(obj_name))
raise Exception("Unable to Create FASTA file from Genome Annotation : {0}".format("".join(traceback.format_exc())))
finally:
fasta_file.close()
fasta_end = datetime.datetime.utcnow()
logger.info("Generating FASTA for {} took {}".format(obj_name, fasta_end - fasta_start))
return output_file
def run(ws_url='https://ci.kbase.us/services/ws/'):
#use a KBase reference genome
genomeref = "ReferenceGenomeAnnotations/kb|g.166819"
#instantiate a new genome annotation API
genome_annotation = GenomeAnnotationAPI(
services={"workspace_service_url": ws_url},
token=os.environ.get('KB_AUTH_TOKEN'),
ref=genomeref)
#pick first locus
gene = 'kb|g.166819.locus.1'
#retriee core exons for locus
exons = get_core_exons(gene, genome_annotation)
print "Core exon(s) found in all mRNAs for gene " + gene
print exons
def run(ws_url='https://ci.kbase.us/services/ws/'):
#an example KBase reference genome
genomeref = "ReferenceGenomeAnnotations/kb|g.166819"
#instantiate a new genome annotation API
genome_annotation = GenomeAnnotationAPI(
services={"workspace_service_url": ws_url},
token=os.environ.get('KB_AUTH_TOKEN'),
ref=genomeref)
#pick the first locus
gene = 'kb|g.166819.locus.1'
#pick the first locus
gffdata = get_gff(gene, genome_annotation)
outfile = 'g.166819.locus.1.gff'
print outfile
with open(outfile, 'w') as f:
f.write(gffdata)
def run(ws_url='https://ci.kbase.us/services/ws/'):
#an example KBase reference genome
genomeref = "ReferenceGenomeAnnotations/kb|g.166819"
#creating a new GenomeAnnotation object
genome_annotation = GenomeAnnotationAPI(
services={"workspace_service_url": ws_url},
token=os.environ.get('KB_AUTH_TOKEN'),
ref=genomeref)
gc = contig_gc(genome_annotation, genomeref)
print gc
outfile = '166819_GC.txt'
with open(outfile, 'w') as f:
f.write(json.dumps(gc))
def run(ws_url='https://ci.kbase.us/services/ws/'):
#an example KBase reference genome
genomeref = "ReferenceGenomeAnnotations/kb|g.166819"
#instantiate a new genome annotation API
genome_annotation = GenomeAnnotationAPI(
services={"workspace_service_url": ws_url},
token=os.environ.get('KB_AUTH_TOKEN'),
ref=genomeref)
genes = [
'kb|g.166819.locus.1', 'kb|g.166819.locus.2', 'kb|g.166819.locus.3'
]
gffdata = ""
for s in genes:
gffdata += get_gff(s, genome_annotation)
outfile = 'g.166819.locus.1_2_3.gff'
print outfile
with open(outfile, 'w') as f:
f.write(gffdata)
