Mercurial > repos > miller-lab > genome_diversity

comparison Population.py @ 17:a3af29edcce2

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded Miller Lab Devshed version a51c894f5bed
author miller-lab
date Fri, 28 Sep 2012 11:57:18 -0400
parents 2c498d40ecde
children 8997f2ca8c7a
comparison
equal deleted inserted replaced
16:be0e2223c531 17:a3af29edcce2
1 #!/usr/bin/env python
2
3 from OrderedDict import OrderedDict
4
5 class Individual(object):
6 __slots__ = ['_column', '_name', '_alias']
7
8 def __init__(self, column, name, alias=None):
9 self._column = column
10 self._name = name
11 self._alias = alias
12
13 @property
14 def column(self):
15 return self._column
16
17 @property
18 def name(self):
19 return self._name if self._alias is None else self._alias
20
21 @property
22 def alias(self):
23 return self._alias
24
25 @alias.setter
26 def alias(self, alias):
27 self._alias = alias
28
29 @property
30 def real_name(self):
31 return self._name
32
33 def __eq__(self, other):
34 return self._column == other._column and self._name == other._name
35
36 def __ne__(self, other):
37 return not self.__eq__(other)
38
39 def __repr__(self):
40 return 'Individual: column={0} name={1} alias={2}'.format(self._column, self._name, self._alias)
41
42
43 class Population(object):
44 def __init__(self, name=None):
45 self._columns = OrderedDict()
46 self._name = name
47
48 @property
49 def name(self):
50 return self._name
51
52 @name.setter
53 def name(self, name):
54 self._name = name
55
56 def add_individual(self, individual, alias=None):
57 if individual.column not in self._columns:
58 self._columns[individual.column] = individual
59 elif self._columns[individual.column] == individual:
60 # should should this be an error?
61 # should we replace the alias using this entry?
62 pass
63 else:
64 raise 'Duplicate column: {0}'.format(individual)
65
66 def is_superset(self, other):
67 for column, other_individual in other._columns.items():
68 our_individual = self._columns.get(column)
69 if our_individual is None or our_individual != other_individual:
70 return False
71 return True
72
73 def is_disjoint(self, other):
74 for column, our_individual in self._columns.items():
75 other_individual = other._columns.get(column)
76 if other_individual is not None and other_individual == our_individual:
77 return False
78 return True
79
80 def column_list(self):
81 return self._columns.keys()
82
83 def individual_with_column(self, column):
84 if column in self._columns:
85 return self._columns[column]
86 return None
87
88 def tag_list(self, delimiter=':'):
89 entries = []
90 for column, individual in self._columns.items():
91 entry = '{0}{1}{2}'.format(column, delimiter, individual.name)
92 entries.append(entry)
93 return entries
94
95 def to_string(self, delimiter=':', separator=' ', replace_names_with=None):
96 entries = []
97 for column, individual in self._columns.items():
98 value = individual.name
99 if replace_names_with is not None:
100 value = replace_names_with
101 entry = '{0}{1}{2}'.format(column, delimiter, value)
102 entries.append(entry)
103 return separator.join(entries)
104
105 def __str__(self):
106 return self.to_string()
107
108 def from_population_file(self, filename):
109 with open(filename) as fh:
110 for line in fh:
111 line = line.rstrip('\r\n')
112 column, name, alias = line.split('\t')
113 alias = alias.strip()
114 individual = Individual(column, name)
115 if alias:
116 individual.alias = alias
117 self.add_individual(individual)
118
119 def from_tag_list(self, tag_list):
120 for tag in tag_list:
121 column, name = tag.split(':')
122 individual = Individual(column, name)
123 self.add_individual(individual)
124
125 def individual_names(self):
126 for column, individual in self._columns.items():
127 yield individual.name
128