def _get_sorted_index(self):
"""Get index of samples indexed by reference and ordered by start pos.
:returns: {ref_name: [sample dicts sorted by start]}
"""
ref_names = defaultdict(list)
for key, f in self.samples:
d = Sample.decode_sample_name(key)
if d is not None:
d['key'] = key
d['filename'] = f
ref_names[d['ref_name']].append(d)
# sort dicts so that refs are in order and within a ref, chunks are in order
ref_names_ordered = OrderedDict()
get_major_minor = lambda x: tuple((int(i) for i in x.split('.')))
# sort by start and -end so that if we have two samples with the same
# start but differrent end points, the longest sample comes first
sorter = lambda x: (get_major_minor(x['start']) + tuple((-i for i in get_major_minor(x['end']))))
for ref_name in sorted(ref_names.keys()):
ref_names[ref_name].sort(key=sorter)
ref_names_ordered[ref_name] = ref_names[ref_name]
return ref_names_ordered
def test_decode_sample_name(self):
expected = [
{
'ref_name': 'contig1',
'start': '0.0',
'end': '4.3'
},
{
'ref_name': 'contig1',
'start': '4.1',
'end': '7.0'
},
]
for expt, sample in zip(expected, self.samples):
self.assertEqual(expt, Sample.decode_sample_name(sample.name))
def _get_sorted_index(self):
"""Get index of samples indexed by reference and ordered by start pos.
:returns: {ref_name: [sample dicts sorted by start]}
"""
ref_names = defaultdict(list)
for key, f in self.samples:
d = Sample.decode_sample_name(key)
if d is not None:
d['key'] = key
d['filename'] = f
ref_names[d['ref_name']].append(d)
# sort dicts so that refs are in order and within a ref, chunks are in order
ref_names_ordered = OrderedDict()
for ref_name in sorted(ref_names.keys()):
sorter = lambda x: float(x['start'])
ref_names[ref_name].sort(key=sorter)
ref_names_ordered[ref_name] = ref_names[ref_name]
return ref_names_ordered
