def mcl_predict(blast_results_file, min_ident, min_cov, evalue, min_length, tmp_dir): if os.path.getsize(blast_results_file) == 0: return dict() blast_df = BlastReader(blast_results_file).df blast_df = blast_df.loc[blast_df['length'] >= min_length] blast_df = blast_df.loc[blast_df['qlen'] <= 400000] blast_df = blast_df.loc[blast_df['qlen'] >= min_length] blast_df = blast_df.loc[blast_df['qcovs'] >= min_cov] blast_df = blast_df.loc[blast_df['qlen'] >= min_length] blast_df = blast_df.reset_index(drop=True) for index, row in blast_df.iterrows(): (seqid, clust_id) = row[1].split('|') blast_df.iloc[index, blast_df.columns.get_loc('sseqid')] = clust_id filtered_blast = os.path.join(tmp_dir, 'filtered_mcl_blast.txt') blast_df.to_csv(filtered_blast, sep='\t', header=False, line_terminator='\n', index=False) mcl_clusters = mcl(filtered_blast, tmp_dir).getclusters() return mcl_clusters
def repetitive_blast(input_fasta, ref_db, min_ident, min_cov, evalue, min_length, tmp_dir, blast_results_file, num_threads=1): blast_runner = BlastRunner(input_fasta, tmp_dir) #blast_runner.makeblastdb(ref_db, 'nucl') blast_runner.run_blast(query_fasta_path=input_fasta, blast_task='megablast', db_path=ref_db, db_type='nucl', min_cov=min_cov, min_ident=min_ident, evalue=evalue, blast_outfile=blast_results_file, num_threads=num_threads) if os.path.getsize(blast_results_file) == 0: return dict() blast_df = BlastReader(blast_results_file).df blast_df = blast_df.loc[blast_df['length'] >= min_length] blast_df = blast_df.loc[blast_df['pident'] >= min_ident] blast_df = blast_df.loc[blast_df['qcovs'] >= min_cov] blast_df = blast_df.loc[blast_df['qcovhsp'] >= 25] blast_df = fixStart(blast_df) blast_df = blast_df.sort_values(['sseqid', 'sstart', 'send', 'bitscore'], ascending=[True, True, True, False]) blast_df = blast_df.reset_index(drop=True) contig_list = dict() for index, row in blast_df.iterrows(): if not row['qseqid'] in contig_list: contig_list[row['qseqid']] = { 'id': row['sseqid'], 'score': row['bitscore'], 'contig_start': row['sstart'], 'contig_end': row['send'] } else: if contig_list[row['qseqid']]['score'] > row['bitscore']: contig_list[row['qseqid']] = { 'id': row['sseqid'], 'score': row['bitscore'], 'contig_start': row['sstart'], 'contig_end': row['send'] } return contig_list
def overhangDetection(self, blast_results_file, min_length=25): if os.path.getsize(blast_results_file) == 0: return dict() blast_df = BlastReader(blast_results_file).df circular_contigs = {} for index, row in blast_df.iterrows(): contig_id_query = row['qseqid'] contig_id_subject = row['sseqid'] contig_start_subject = row['sstart'] contig_end_subject = row['send'] contig_start_query = row['qstart'] contig_end_query = row['qend'] contig_length = row['qlen'] mid_point = int(contig_length / 2) length = row['length'] if contig_id_query != contig_id_subject: continue if contig_start_query != 1 or length < min_length: continue if contig_start_query == contig_start_subject and contig_end_query == contig_end_subject: continue if (contig_start_query == 1 and contig_end_query == mid_point) or \ (contig_start_query == mid_point+1 and contig_end_query == contig_length): circular_contigs[ contig_id_query] = 'Circular: Complete concatemer' elif contig_start_query == 1 and contig_end_subject == contig_length: circular_contigs[ contig_id_query] = 'Circular: Overlap {} bp'.format(length) return circular_contigs
def overhangDetection(self, blast_results_file, logging, min_length=25): if os.path.getsize(blast_results_file) == 0: return dict() blast_df = BlastReader(blast_results_file, logging).df.sort_values( ['qseqid', 'qstart', 'qend', 'bitscore'], ascending=[True, True, True, False]) circular_contigs = {} for index, row in blast_df.iterrows(): contig_id_query = row['qseqid'] contig_id_subject = row['sseqid'] contig_start_subject = int(row['sstart']) contig_end_subject = int(row['send']) contig_start_query = int(row['qstart']) contig_end_query = int(row['qend']) contig_length = int(row['qlen']) length = int(row['length']) if contig_id_query != contig_id_subject and contig_id_subject != "ref|{}|".format( contig_id_query): continue if contig_start_query != 1 or length < min_length: continue if contig_start_query == contig_start_subject and contig_end_query == contig_end_subject: continue if contig_start_query == 1 and contig_end_subject == contig_length: circular_contigs[ contig_id_query] = 'Circular: Overlap {} bp'.format(length) return circular_contigs
def contig_blast_group(blast_results_file, overlap_threshold): if os.path.getsize(blast_results_file) == 0: return dict() blast_df = BlastReader(blast_results_file).df blast_df = blast_df.sort_values(['sseqid', 'sstart', 'send', 'bitscore'], ascending=[True, True, True, False]) blast_df = filter_overlaping_records(blast_df, overlap_threshold, 'sseqid', 'sstart', 'send', 'bitscore') size = str(len(blast_df)) prev_size = 0 while size != prev_size: blast_df = filter_overlaping_records(blast_df, overlap_threshold, 'sseqid', 'sstart', 'send', 'bitscore') prev_size = size size = str(len(blast_df)) cluster_scores = dict() groups = dict() hits = dict() contigs = dict() for index, row in blast_df.iterrows(): query = row['qseqid'] pID, clust_id = row['sseqid'].split('|') score = row['bitscore'] pLen = row['slen'] contig_id = row['qseqid'] if not pID in hits: hits[pID] = { 'score': 0, 'length': pLen, 'covered_bases': 0, 'clust_id': clust_id } if not clust_id in cluster_scores: cluster_scores[clust_id] = score elif score > cluster_scores[clust_id]: cluster_scores[clust_id] = score if not clust_id in groups: groups[clust_id] = dict() if not query in groups[clust_id]: groups[clust_id][query] = dict() if not contig_id in contigs: contigs[contig_id] = dict() if not clust_id in contigs[contig_id]: contigs[contig_id][clust_id] = 0 if contigs[contig_id][clust_id] < score: contigs[contig_id][clust_id] = score groups[clust_id][query][contig_id] = score hits[pID]['score'] += score hits[pID]['covered_bases'] += score sorted_d = OrderedDict( sorted(iter(list(cluster_scores.items())), key=lambda x: x[1], reverse=True)) for clust_id in sorted_d: score = sorted_d[clust_id] for contig_id in contigs: if clust_id in contigs[contig_id]: contigs[contig_id] = {clust_id: contigs[contig_id][clust_id]} return contigs