qN26Ds9a

1@hug.post('/search', response_headers={'Access-Control-Allow-Origin': '*'},)
2def search(seq: hug.types.text,
3           threshold: hug.types.float_number = 0.4,  # bgsi recommendation 
4           score: hug.types.smart_boolean = False):
5    """
6    Search a DNA sequence against the index of the human_gut using BIGSI
7    BIGSI stands for: BIGSIs–BItsliced Genomic Signature Indexes.
8    :seq: The fasta sequence to be searched on the index.
9    :param print_cols: A flag used to print the columns to the console
10        (default is False)
11    :returns: a dictionary with the original query, threshold and results.
12        Thee results is a list of objects with the following structure:
13        {
14            bigsi: {
15                sample_name
16                percent_kmers_found
17                num_kmers
18                num_kmers_found
19            },
20            mgnify: {
21                id,
22                attributes: -- MGnify genomes API data  --
23            }
24        } 
25    :rtype: dict
26    """
27    fasta_seq = _clean_fasta(seq)
28
29    if seq.count('>') > 1:
30        raise falcon.HTTPBadRequest('seq', 'Please provide just one DNA fasta sequence.')
31
32    if not re.match('^[ATGCRYMKSWHBVDN\s]+$', fasta_seq, re.IGNORECASE):
33        raise falcon.HTTPBadRequest('seq', 'The sequence doesn\'t appear to be a DNA sequence')
34
35    if len(fasta_seq) > MAX_LEN or len(fasta_seq) < MIN_LEN:
36        raise falcon.HTTPBadRequest('seq', f'The sequence should be longer that {MIN_LEN} and ' + 
37                                    'shorter than {MAX_LEN}pb')
38
39    best_matches = bigsi.search(fasta_seq, threshold, score)
40    best_matches = sorted(best_matches, key=itemgetter('percent_kmers_found'), reverse=True)
41
42    # merge the data from the DB
43    storage = db.DB()
44    storage.open(MGNIFY_CACHE_PATH, None, db.DB_HASH, db.DB_READ_COMMITTED)
45    
46    results = []
47
48    for hit in best_matches:
49        key = hit.get('sample_name')
50        data = storage.get(key.encode('utf-8'))
51        results.append({
52            'mgnify': json.loads(data) if data else {}, 
53            'bigsi': hit 
54        })
55
56    storage.close()
57
58    return {
59        'query': fasta_seq,
60        'threshold': threshold,
61        'results': results
62    }