def load_protein_configuration(protein_id, ref_species_dict = None):
'''
Loads the data from a description file, and calls the containers generating functions to create basic objects.
@param protein_id: id of a single protein
'''
if not check_status_file_no_alignment(protein_id):
return False
if ref_species_dict is None:
ref_species_dict = FileUtilities.get_reference_species_dictionary()
logger = Logger.Instance()
containers_logger = logger.get_logger('containers')
data_map_container = DataMapContainer.Instance()
protein_container = ProteinContainer.Instance()
gene_container = GeneContainer.Instance()
transcript_container = TranscriptContainer.Instance()
ens_exon_container = EnsemblExonContainer.Instance()
(known_proteins, abinitio_proteins) = DescriptionParser().parse_description_file_general_info(protein_id)
for species_data in known_proteins:
(species_name,
spec_protein_id,
spec_gene_id,
spec_transcript_id,
location_type,
assembly,
location_id,
seq_begin,
seq_end,
strand) = species_data
ab_initio = False
# data map
data_map_key = (protein_id, species_name)
data_map = DataMap(spec_protein_id, spec_transcript_id,
spec_gene_id, data_map_key, location_type, assembly,
location_id, strand, seq_begin, seq_end, ab_initio)
try:
data_map_container.add(data_map_key, data_map)
except (KeyError, TypeError), e:
containers_logger.error("{0}, {1}, {2}, error adding to datamap".format(protein_id, species_name, e.args[0]))
# everything else - protein, transcript, gene, ensembl exons
protein = Protein(spec_protein_id, data_map_key, ref_species_dict[species_name])
gene = Gene(spec_gene_id, data_map_key, ref_species_dict[species_name])
transcript = Transcript(spec_transcript_id, data_map_key, ref_species_dict[species_name])
ens_exons = EnsemblExons(data_map_key, ref_species_dict[species_name])
try:
ens_exons.load_exons()
except (Exception), e:
containers_logger.error("{0}, {1}, {2}, error loading exons".format(protein_id, species_name, e.args[0]))
def load_exon_configuration (ref_protein_id, ref_species_dict, exon_type):
'''
Load exons of a particular type for all available species
@param ref_protein_id: referent protein id
@param exon_type: exon_type: ensembl, genewise, blatn, tblastn, sw_gene, sw_exon
'''
dc = DescriptionParser()
exon_container = ExonContainer.Instance()
ens_exon_container = EnsemblExonContainer.Instance()
logger = Logger.Instance()
containers_logger = logger.get_logger('containers')
if exon_type == "ensembl" or exon_type == "genewise":
if not check_status_file_no_alignment(ref_protein_id):
containers_logger.info ("{0},exon_type:{1},check status file -> failed".format(ref_protein_id, exon_type))
return False
else:
if not check_status_file(ref_protein_id):
containers_logger.info ("{0},exon_type:{1},check status file -> failed".format(ref_protein_id, exon_type))
return False
if not ref_species_dict:
ref_species_dict = FileUtilities.get_reference_species_dictionary()
(known_species, abinitio_species) = dc.get_separated_species(ref_protein_id)
for species in known_species:
ref_species = ref_species_dict[species]
if exon_type != "genewise":
if exon_type == "ensembl":
exons = EnsemblExons ((ref_protein_id, species), ref_species)
try:
exon_dict = exons.load_exons()
except Exception, e:
containers_logger.error("{0},{1},{2},error loading exons".format(ref_protein_id, species, exon_type))
continue
else:
exons = Exons((ref_protein_id, species), ref_species, exon_type)
try:
exon_dict = exons.load_exons()
except Exception, e:
containers_logger.error("{0},{1},{2},error loading exons".format(ref_protein_id, species, exon_type))
continue
if not exon_dict:
continue
if (exon_type != "ensembl"):
exons.set_exon_ordinals()
data_map_key = [ref_protein_id, species]
exon_container.add(exon_type, data_map_key, exons)
