def test_define_groups(self):
me = Analyze()
me.taxdump = taxdump_from_text(taxdump_proteo)
add_children(me.taxdump)
me.groups = {}
# user defined groups:
# self: genera Escherichia and Shigella
# close: family Enterobacteriaceae
me.groups = {}
me.self_tax = '561,620'
me.close_tax = '543'
me.define_groups()
self.assertListEqual(me.self_tax, ['561', '620'])
exp = {'561', '562', '585056', '1038927', '2580236', '620', '622'}
self.assertSetEqual(me.groups['self'], exp)
self.assertListEqual(me.close_tax, ['543'])
exp = {'543', '548', '570'}
self.assertSetEqual(me.groups['close'], exp)
# auto-infer groups
me.self_tax = {}
me.close_tax = {}
me.lca = '562' # all inputs are E. coli
me.self_rank = 'genus' # but we want to raise self to genus
me.close_size = 2 # close group must be this big or bigger
me.define_groups()
self.assertListEqual(me.self_tax, ['561'])
exp = {'561', '562', '585056', '1038927', '2580236'}
self.assertSetEqual(me.groups['self'], exp)
self.assertListEqual(me.close_tax, ['543'])
exp = {'543', '548', '570', '620', '622'}
self.assertSetEqual(me.groups['close'], exp)
def test_infer_close_group(self):
me = Analyze()
me.taxdump = taxdump_from_text(taxdump_proteo)
add_children(me.taxdump)
me.groups = {}
# close group is parent of LCA of self group
me.self_tax = ['562'] # E. coli
me.groups['self'] = set(['562'] + get_descendants('562', me.taxdump))
me.close_tax = None
me.close_size = None
me.infer_close_group()
self.assertListEqual(me.close_tax, ['561']) # Escherichia
self.assertSetEqual(me.groups['close'], {'561', '2580236'})
# close group must have at least 5 taxa
me.close_tax = None
me.groups['close'] = None
me.close_size = 5
me.infer_close_group()
self.assertListEqual(me.close_tax, ['543']) # Enterobacteriaceae
exp = {'543', '620', '622', '570', '548', '561', '2580236'}
self.assertSetEqual(me.groups['close'], exp)
# close group is LCA of multiple self groups
me.self_tax = ['561', '620'] # Escherichia and Shigella
me.groups['self'] = set().union(*[[x] + get_descendants(x, me.taxdump)
for x in me.self_tax])
me.close_tax = None
me.groups['close'] = None
me.close_size = None
me.infer_close_group()
self.assertListEqual(me.close_tax, ['543']) # Enterobacteriaceae
exp = {'543', '570', '548'}
self.assertSetEqual(me.groups['close'], exp)
def test_calc_scores(self):
columns = ('id', 'taxid', 'score')
# helper for making hit table
def _hits_df(data):
return pd.DataFrame(data, columns=columns).set_index('id')
me = Analyze()
me.taxdump = taxdump_from_text(taxdump_proteo)
add_children(me.taxdump)
me.groups = {
'self': {'561', '562', '585056'},
'close': {'543', '91347', '1236'}
}
me.data = {
'S1': [{
'score': 100,
'hits': _hits_df((('P1', '561', 100), ('P2', '562', 95)))
}, {
'score': 90,
'hits': _hits_df((('P3', '561', 81), ('P4', '543', 72)))
}],
'S2': [{
'score':
96,
'hits':
_hits_df(
(('P5', '561', 90), ('P6', '543', 84), ('P7', '620', 66)))
}]
}
me.weighted = True
me.match_th = 0.9
me.calc_scores()
# helper for get scores
def _prot_scores(prot):
return [prot[x] for x in ('self', 'close', 'distal')]
s1_1 = me.data['S1'][0]
self.assertListEqual(s1_1['hits']['group'].tolist(), ['self', 'self'])
self.assertListEqual(_prot_scores(s1_1), [1.95, 0.0, 0.0])
self.assertEqual(s1_1['match'], '0')
s1_2 = me.data['S1'][1]
self.assertListEqual(s1_2['hits']['group'].tolist(), ['self', 'close'])
self.assertListEqual(_prot_scores(s1_2), [0.9, 0.8, 0.0])
self.assertEqual(s1_2['match'], '0')
s2_1 = me.data['S2'][0]
self.assertListEqual(s2_1['hits']['group'].tolist(),
['self', 'close', 'distal'])
self.assertListEqual(_prot_scores(s2_1), [0.9375, 0.875, 0.6875])
self.assertEqual(s2_1['match'], '620')
