def load(datadir, create_session=None):
'''
nr_loaded = load(datadir, create_session={backend.create_session})
Load ENSEMBL FASTA file into database
Parameters
----------
datadir : str
Directory containing the FASTA file
create_session : callable, optional
a callable object that returns an sqlalchemy session
Returns
-------
nr_loaded : integer
Nr. of entries loaded
'''
from waldo.backend import call_create_session
session = call_create_session(create_session)
inputfilename = glob.glob(path.join(datadir, 'Mus_musculus.*.pep.all.fa.gz'))[0]
filename = path.join(inputfilename)
nr_loaded = 0
for seq in fasta.read(filename):
htokens = seq.header.split()
peptide = htokens[0]
gene = htokens[3]
assert gene.startswith('gene:'), 'waldo.sequences.load'
gene = gene[len('gene:'):]
session.add(
Translation(
'ensembl:gene_id',
gene,
'ensembl:peptide_id',
peptide))
aaseq = seq.sequence
seq = models.EnsemblSequence(peptide, aaseq)
session.add(seq)
session.commit()
nr_loaded += 1
return nr_loaded
def test_read():
seqs = list(fasta.read(path_to_testfile('test.fasta')))
assert len(seqs) == 2
assert seqs[1].header.find('gene:ENSMUSG00000064345')
assert seqs[1].header[-1] != '\n'
assert seqs[1].header[0] != '>'