def test_make_snv_key():
test_data = (
# chrom, pos, ref, alt
(('1', '211', 'A', 'C'), ('1', '211', 'A', 'c')),
(('X', '2012', 'T', 'G'), ('X', '2012', 't', 'g')),
)
for attributes, equivalent_attributes in test_data:
result_1 = genomic_mappings.make_snv_key(*attributes)
result_2 = genomic_mappings.make_snv_key(*equivalent_attributes)
assert result_1 == result_2
def test_mappings():
"""This is a simple inclusion test for genome -> proteme mutation mappings.
Knowing the data, we demand the items from the right side (of test data)
to be in the results of queries specified on the left side.
"""
test_data = (
# (chrom, dna_pos, dna_ref, dna_alt), (name, pos, ref, alt)
(('17', '7572934', 'G', 'A'), ('TP53', 353, 'S', 'L')),
(('17', '19282215', 't', 'a'), ('MAPK7', 1, 'M', 'K')),
(('21', '40547520', 'g', 'a'), ('PSMG1', 283, 'T', 'I')),
(('9', '125616157', 't', 'a'), ('RC3H2', 1064, 'Y', 'F')),
(('11', '120198175', 'g', 'a'), ('TMEM136', 31, 'V', 'M')),
(('10', '81838457', 't', 'a'), ('TMEM254', 1, 'M', 'K')),
(('13', '111267940', 't', 'a'), ('CARKD', 1, 'M', 'K')),
(('6', '30539266', 't', 'a'), ('ABCF1', 1, 'M', 'K')),
(('6', '36765430', 'g', 'a'), ('CPNE5', 140, 'L', 'F')),
(('12', '123464753', 't', 'a'), ('ARL6IP4', 1, 'M', 'K')),
)
for genomic_data, protein_data in test_data:
snv = make_snv_key(*genomic_data)
items = [decode_csv(item) for item in bdb[snv]]
for item in items:
retrieved_data = (Protein.query.get(item['protein_id']).gene.name,
item['pos'], item['ref'], item['alt'])
if retrieved_data == protein_data:
break
else:
raise Exception(retrieved_data, protein_data)
def test_genome_proteome_mappings(self):
mappings_filename, gene, proteins = create_test_data()
broken_sequences = import_genome_proteome_mappings(
proteins, path.dirname(mappings_filename),
path.basename(mappings_filename))
# in some cases it is needed to reload bdb after import
bdb.reload()
assert not bdb[make_snv_key('1', 19282216, 'G', 'A')]
assert bdb[make_snv_key('17', 19282216, 'G', 'A')]
assert set(broken_sequences.keys()) == {'NM_002749'}
assert [('NM_002749', 'L', 'A', '5', 'Q')
] in list(broken_sequences.values())
def import_genome_proteome_mappings(
proteins,
mappings_dir='data/200616/all_variants/playground',
mappings_file_pattern='annot_*.txt.gz',
bdb_dir=''):
print('Importing mappings:')
chromosomes = get_human_chromosomes()
broken_seq = defaultdict(list)
bdb.reset()
bdb.close()
path = current_app.config['BDB_DNA_TO_PROTEIN_PATH']
if bdb_dir:
path = bdb_dir + '/' + basename(path)
bdb.open(path, cache_size=20480 * 8 * 8 * 8 * 8)
for line in read_from_gz_files(mappings_dir, mappings_file_pattern):
try:
chrom, pos, ref, alt, prot = line.rstrip().split('\t')
except ValueError as e:
print(e, line)
continue
assert chrom.startswith('chr')
chrom = chrom[3:]
assert chrom in chromosomes
ref = ref.rstrip()
# new Coding Sequence Variants to be added to those already
# mapped from given `snv` (Single Nucleotide Variation)
for dest in filter(bool, prot.split(',')):
try:
name, refseq, exon, cdna_mut, prot_mut = dest.split(':')
except ValueError as e:
print(e, line)
continue
assert refseq.startswith('NM_')
# refseq = int(refseq[3:])
# name and refseq are redundant with respect one to another
assert exon.startswith('exon')
exon = exon[4:]
assert cdna_mut.startswith('c')
try:
cdna_ref, cdna_pos, cdna_alt = decode_mutation(cdna_mut)
except ValueError as e:
print(e, line)
continue
try:
strand = determine_strand(ref, cdna_ref, alt, cdna_alt)
except DataInconsistencyError as e:
print(e, line)
continue
assert prot_mut.startswith('p')
# we can check here if a given reference nuc is consistent
# with the reference amino acid. For example cytosine in
# reference implies that there should't be a methionine,
# glutamic acid, lysine nor arginine. The same applies to
# alternative nuc/aa and their combinations (having
# references (nuc, aa): (G, K) and alt nuc C defines that
# the alt aa has to be Asparagine (N) - no other is valid).
# Note: it could be used to compress the data in memory too
aa_ref, aa_pos, aa_alt = decode_mutation(prot_mut)
try:
# try to get it from cache (`proteins` dictionary)
protein = proteins[refseq]
except KeyError:
continue
assert aa_pos == (int(cdna_pos) - 1) // 3 + 1
broken_sequence_tuple = is_sequence_broken(protein, aa_pos, aa_ref,
aa_alt)
if broken_sequence_tuple:
broken_seq[refseq].append(broken_sequence_tuple)
continue
is_ptm_related = protein.has_sites_in_range(aa_pos - 7, aa_pos + 7)
snv = make_snv_key(chrom, pos, cdna_ref, cdna_alt)
# add new item, emulating set update
item = encode_csv(strand, aa_ref, aa_alt, cdna_pos, exon,
protein.id, is_ptm_related)
bdb.add(snv, item)
return broken_seq
