def generate_variabile_aas_table(self):
if self.skip_SNV_profiling or not self.profile_AA_frequencies:
# there is nothing to generate really..
self.run.info('AA_frequencies_table', False, quiet=True)
return
self.progress.new('Computing AA frequencies at variable positions')
variable_aas_table = dbops.TableForAAFrequencies(
self.profile_db_path, progress=self.progress)
aa_frequencies = bamops.AAFrequencies()
codons_in_genes_to_profile_AA_frequencies_dict = {}
for gene_call_id, codon_order in self.codons_in_genes_to_profile_AA_frequencies:
if gene_call_id not in codons_in_genes_to_profile_AA_frequencies_dict:
codons_in_genes_to_profile_AA_frequencies_dict[
gene_call_id] = set([])
codons_in_genes_to_profile_AA_frequencies_dict[gene_call_id].add(
codon_order)
gene_caller_ids_to_profile = codons_in_genes_to_profile_AA_frequencies_dict.keys(
)
num_gene_caller_ids_to_profile = len(gene_caller_ids_to_profile)
for i in range(0, len(gene_caller_ids_to_profile)):
gene_caller_id = gene_caller_ids_to_profile[i]
codons_to_profile = codons_in_genes_to_profile_AA_frequencies_dict[
gene_caller_id]
self.progress.update("Working on gene caller id '%d' (%d of %d) w/ %d codons of interest" \
% (gene_caller_id, i + 1, num_gene_caller_ids_to_profile, len(codons_to_profile)))
gene_call = self.genes_in_contigs_dict[gene_caller_id]
contig_name = gene_call['contig']
aa_frequencies_dict = aa_frequencies.process_gene_call(
self.bam, gene_call,
self.contig_sequences[contig_name]['sequence'],
codons_to_profile)
for codon_order in aa_frequencies_dict:
e = aa_frequencies_dict[codon_order]
db_entry = {
'sample_id': self.sample_id,
'corresponding_gene_call': gene_caller_id
}
db_entry['reference'] = e['reference']
db_entry['coverage'] = e['coverage']
db_entry['departure_from_reference'] = e[
'departure_from_reference']
db_entry['codon_order_in_gene'] = codon_order
for aa in constants.codon_to_AA.values():
db_entry[aa] = e['frequencies'][aa]
variable_aas_table.append(db_entry)
variable_aas_table.store()
self.progress.end()
self.run.info('AA_frequencies_table', True, quiet=True)
def generate_variabile_aas_table(self):
if self.skip_SNV_profiling or not self.profile_AA_frequencies:
return
variable_aas_table = dbops.TableForAAFrequencies(
self.profile_db_path, progress=self.progress)
aa_frequencies = bamops.AAFrequencies()
codons_in_genes_to_profile_AA_frequencies_dict = {}
for gene_call_id, codon_order in self.codons_in_genes_to_profile_AA_frequencies:
if gene_call_id not in codons_in_genes_to_profile_AA_frequencies_dict:
codons_in_genes_to_profile_AA_frequencies_dict[
gene_call_id] = set([])
codons_in_genes_to_profile_AA_frequencies_dict[gene_call_id].add(
codon_order)
gene_caller_ids_to_profile = list(
codons_in_genes_to_profile_AA_frequencies_dict.keys())
for i in range(0, len(gene_caller_ids_to_profile)):
gene_caller_id = gene_caller_ids_to_profile[i]
codons_to_profile = codons_in_genes_to_profile_AA_frequencies_dict[
gene_caller_id]
gene_call = self.genes_in_contigs_dict[gene_caller_id]
contig_name = gene_call['contig']
aa_frequencies_dict = aa_frequencies.process_gene_call(
self.bam, gene_call,
self.contig_sequences[contig_name]['sequence'],
codons_to_profile)
for codon_order in aa_frequencies_dict:
e = aa_frequencies_dict[codon_order]
db_entry = {
'sample_id': self.sample_id,
'corresponding_gene_call': gene_caller_id
}
db_entry['reference'] = e['reference']
db_entry['coverage'] = e['coverage']
db_entry['departure_from_reference'] = e[
'departure_from_reference']
db_entry['codon_order_in_gene'] = codon_order
for aa in list(constants.codon_to_AA.values()):
db_entry[aa] = e['frequencies'][aa]
variable_aas_table.append(db_entry)
variable_aas_table.store()
# clear contents of set
self.codons_in_genes_to_profile_AA_frequencies.clear()
def merge_variable_aas_tables(self):
variable_aas_table = dbops.TableForAAFrequencies(
self.merged_profile_db_path, progress=self.progress)
for input_profile_db_path in self.profile_dbs_info_dict:
sample_profile_db = dbops.ProfileDatabase(input_profile_db_path,
quiet=True)
sample_variable_aas_table = sample_profile_db.db.get_table_as_list_of_tuples(
tables.variable_aas_table_name,
tables.variable_aas_table_structure)
sample_profile_db.disconnect()
for tpl in sample_variable_aas_table:
entry = tuple([
variable_aas_table.next_id(tables.variable_aas_table_name)
] + list(tpl[1:]))
variable_aas_table.db_entries.append(entry)
variable_aas_table.store()
def merge_variable_aas_tables(self):
self.is_all_samples_have_it('AA_frequencies_table')
variable_aas_table = dbops.TableForAAFrequencies(
self.profile_db_path, progress=self.progress)
for runinfo in self.input_runinfo_dicts.values():
sample_profile_db = dbops.ProfileDatabase(runinfo['profile_db'],
quiet=True)
sample_variable_aas_table = sample_profile_db.db.get_table_as_list_of_tuples(
tables.variable_aas_table_name,
tables.variable_aas_table_structure)
sample_profile_db.disconnect()
for tpl in sample_variable_aas_table:
entry = tuple([
variable_aas_table.next_id(tables.variable_aas_table_name)
] + list(tpl[1:]))
variable_aas_table.db_entries.append(entry)
variable_aas_table.store()