Completed
Push — master ( c2640a...0b9f50 )
by
unknown
54s
created

MgrNtGOEAs   A

Complexity

Total Complexity 24

Size/Duplication

Total Lines 69
Duplicated Lines 0 %

Importance

Changes 1
Bugs 0 Features 0
Metric Value
c 1
b 0
f 0
dl 0
loc 69
rs 10
wmc 24

6 Methods

Rating   Name   Duplication   Size   Complexity  
A _get_field_values() 0 7 4
A get_goea_nts_prt() 0 8 3
F get_goea_nts_all() 0 31 11
A __init__() 0 2 1
A get_study_items() 0 6 2
A _get_fieldnames() 0 7 3
1
"""Manage GOATOOLS GOEA namedtuples."""
2
3
__copyright__ = "Copyright (C) 2010-2018, H Tang et al., All rights reserved."
4
__author__ = "DV Klopfenstein"
5
6
import collections as cx
7
8
9
def get_study_items(goea_results):
10
    """Get all study items found in a GOATOOLS GOEA (e.g., geneids)."""
11
    return MgrNtGOEAs(goea_results).get_study_items()
12
13
def get_goea_nts_prt(goea_results, **kws):
14
    """Get namedtuples containing user-specified (or default) data from GOATOOLS GOEA results."""
15
    return MgrNtGOEAs(goea_results).get_goea_nts_prt(**kws)
16
17
18
class MgrNtGOEAs(object):
19
    """Manage GOATOOLS GOEA namedtuples."""
20
21
    def __init__(self, goea_results):
22
        self.goea_results = goea_results
23
24
    def get_study_items(self):
25
        """Get all study items (e.g., geneids)."""
26
        study_items = set()
27
        for rec in self.goea_results:
28
            study_items |= rec.study_items
29
        return study_items
30
31
    def get_goea_nts_prt(self, fldnames=None, **usr_kws):
32
        """Return list of namedtuples removing fields which are redundant or verbose."""
33
        kws = usr_kws.copy()
34
        if 'not_fldnames' not in kws:
35
            kws['not_fldnames'] = ['goterm', 'parents', 'children', 'id']
36
        if 'rpt_fmt' not in kws:
37
            kws['rpt_fmt'] = True
38
        return self.get_goea_nts_all(fldnames, **kws)
39
40
    def get_goea_nts_all(self, fldnames=None, **kws):
41
        """Get namedtuples containing user-specified (or default) data from GOEA results.
42
43
            Reformats data from GOEnrichmentRecord objects into lists of
44
            namedtuples so the generic table writers may be used.
45
        """
46
        # kws: prt_if indent itemid2name(study_items)
47
        data_nts = [] # A list of namedtuples containing GOEA results
48
        if not self.goea_results:
49
            return data_nts
50
        keep_if = kws.get('keep_if', None)
51
        rpt_fmt = kws.get('rpt_fmt', False)
52
        indent = kws.get('indent', False)
53
        # I. FIELD (column) NAMES
54
        not_fldnames = kws.get('not_fldnames', None)
55
        if fldnames is None:
56
            fldnames = self._get_fieldnames(self.goea_results[0])
57
        # Ia. Explicitly exclude specific fields from named tuple
58
        if not_fldnames is not None:
59
            fldnames = [f for f in fldnames if f not in not_fldnames]
60
        nttyp = cx.namedtuple("NtGoeaResults", " ".join(fldnames))
61
        goid_idx = fldnames.index("GO") if 'GO' in fldnames else None
62
        # II. Loop through GOEA results stored in a GOEnrichmentRecord object
63
        for goerec in self.goea_results:
64
            vals = self._get_field_values(goerec, fldnames, rpt_fmt, kws.get('itemid2name', None))
65
            if indent:
66
                vals[goid_idx] = "".join([goerec.get_indent_dots(), vals[goid_idx]])
67
            ntobj = nttyp._make(vals)
68
            if keep_if is None or keep_if(goerec):
69
                data_nts.append(ntobj)
70
        return data_nts
71
72
    @staticmethod
73
    def _get_field_values(item, fldnames, rpt_fmt=None, itemid2name=None):
74
        """Return fieldnames and values of either a namedtuple or GOEnrichmentRecord."""
75
        if hasattr(item, "_fldsdefprt"): # Is a GOEnrichmentRecord
76
            return item.get_field_values(fldnames, rpt_fmt, itemid2name)
77
        if hasattr(item, "_fields"): # Is a namedtuple
78
            return [getattr(item, f) for f in fldnames]
79
80
    @staticmethod
81
    def _get_fieldnames(item):
82
        """Return fieldnames of either a namedtuple or GOEnrichmentRecord."""
83
        if hasattr(item, "_fldsdefprt"): # Is a GOEnrichmentRecord
84
            return item.get_prtflds_all()
85
        if hasattr(item, "_fields"): # Is a namedtuple
86
            return item._fields
87
88
# Copyright (C) 2010-2018, H Tang et al., All rights reserved.
89