allelefinder.py 9.09 KB
Newer Older
jhoogenboom's avatar
jhoogenboom committed
1
2
#!/usr/bin/env python
"""
3
4
Find true alleles in reference samples and detect possible
contaminations.
jhoogenboom's avatar
jhoogenboom committed
5
6
7

In each sample, the sequences with the highest read counts of each
marker are called alleles, with a user-defined maximum number of alleles
jhoogenboom's avatar
jhoogenboom committed
8
per marker.  The allele balance is kept within given bounds.  If the
jhoogenboom's avatar
jhoogenboom committed
9
10
11
highest non-allelic sequence exceeds a given limit, no alleles are
called for this marker.  If this happens for multiple markers in one
sample, no alleles are called for this sample at all.
jhoogenboom's avatar
jhoogenboom committed
12
13
14
"""
import argparse

15
from ..lib import pos_int_arg, add_input_output_args, get_input_output_files, \
16
                  ensure_sequence_format, get_sample_data, parse_library, \
17
                  add_sequence_format_args
jhoogenboom's avatar
jhoogenboom committed
18
19
20
21

__version__ = "0.1dev"


22
23
24
25
26
27
28
29
30
# Default values for parameters are specified below.

# Default minimum number of reads required for the highest allele.
# This value can be overridden by the -n command line option.
_DEF_MIN_READS = 50

# Default minimum number of reads required for an allele to be called,
# as a percentage of the number of reads of the highest allele.
# This value can be overridden by the -m command line option.
jhoogenboom's avatar
jhoogenboom committed
31
_DEF_MIN_ALLELE_PCT = 30.0
32
33
34
35
36
37

# Default maximum amount of noise to allow, as a percentage of the
# number of reads of the highest allele of each marker.  If any noise
# (i.e., non-allelic sequences) above this threshold are detected, the
# sample is considered 'noisy' for this marker.
# This value can be overridden by the -M command line option.
jhoogenboom's avatar
jhoogenboom committed
38
_DEF_MAX_NOISE_PCT = 10.0
39
40
41
42
43
44

# Default maximum number of noisy markers allowed per sample.
# This value can be overridden by the -x command line option.
_DEF_MAX_NOISY = 2


45
46
47
def find_alleles(samples_in, outfile, reportfile, min_reads, min_allele_pct,
                 max_noise_pct, max_alleles, max_noisy, stuttermark_column,
                 seqformat, library):
48
    library = parse_library(library) if library is not None else {}
49

jhoogenboom's avatar
jhoogenboom committed
50
    outfile.write("\t".join(["sample", "marker", "total", "allele"]) + "\n")
51
52
    allelelist = {}
    get_sample_data(
53
        samples_in,
54
55
56
        lambda tag, data: find_alleles_sample(
            data if stuttermark_column is None
                 else {key: data[key] for key in allelelist[tag]},
jhoogenboom's avatar
jhoogenboom committed
57
            outfile, reportfile, tag, min_reads, min_allele_pct, max_noise_pct,
58
59
60
            max_alleles, max_noisy, seqformat, library),
        allelelist,
        stuttermark_column)
61
62
63
#find_alleles


jhoogenboom's avatar
jhoogenboom committed
64
65
66
def find_alleles_sample(data, outfile, reportfile, tag, min_reads,
                        min_allele_pct, max_noise_pct, max_alleles, max_noisy,
                        seqformat, library):
jhoogenboom's avatar
jhoogenboom committed
67
68
69
70
    top_noise = {}
    top_allele = {}
    alleles = {}
    for marker, allele in data:
71
        reads = sum(data[marker, allele])
jhoogenboom's avatar
jhoogenboom committed
72

73
74
75
76
77
78
        if marker not in alleles:
            alleles[marker] = {allele: reads}
            top_allele[marker] = reads
            top_noise[marker] = ["-", 0]
        else:
            if reads > top_allele[marker]:
jhoogenboom's avatar
jhoogenboom committed
79
                # New highest allele!
80
81
82
83
84
85
86
87
88
89
                top_allele[marker] = reads
                for allelex in alleles[marker].keys():
                    if (alleles[marker][allelex] <
                            top_allele[marker] * (min_allele_pct/100.)):
                        if alleles[marker][allelex] > top_noise[marker][1]:
                            top_noise[marker] = [
                                allelex, alleles[marker][allelex]]
                        del alleles[marker][allelex]
                alleles[marker][allele] = reads
            elif reads >= top_allele[marker]*(min_allele_pct/100.):
jhoogenboom's avatar
jhoogenboom committed
90
91
                # New secundary allele!
                alleles[marker][allele] = reads
92
            elif reads >= top_noise[marker][1]:
jhoogenboom's avatar
jhoogenboom committed
93
                # New highest noise!
94
                top_noise[marker] = [allele, reads]
jhoogenboom's avatar
jhoogenboom committed
95

96
97
    # Find and eliminate noisy markers in this sample first.
    noisy_markers = 0
jhoogenboom's avatar
jhoogenboom committed
98
    for marker in alleles:
99
        if top_allele[marker] < min_reads:
jhoogenboom's avatar
jhoogenboom committed
100
101
102
103
            reportfile.write(
                "Sample %s is not suitable for marker %s:\n"
                "highest allele has only %i reads\n\n" %
                    (tag, marker, top_allele[marker]))
104
105
            alleles[marker] = {}
            continue
106
107
        expect = get_max_expected_alleles(max_alleles, marker, library)
        if len(alleles[marker]) > expect:
jhoogenboom's avatar
jhoogenboom committed
108
109
            allele_order = sorted(alleles[marker],
                                  key=lambda x: -alleles[marker][x])
110
111
            top_noise[marker] = [allele_order[expect],
                alleles[marker][allele_order[expect]]]
jhoogenboom's avatar
jhoogenboom committed
112
            alleles[marker] = {x: alleles[marker][x]
113
                               for x in allele_order[:expect]}
114
        if top_noise[marker][1] > top_allele[marker]*(max_noise_pct/100.):
jhoogenboom's avatar
jhoogenboom committed
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
            reportfile.write(
                "Sample %s is not suitable for marker %s:\n"
                "highest non-allele is %.1f%% of the highest allele\n" %
                (tag, marker, 100.*top_noise[marker][1]/top_allele[marker]))
            for allele in sorted(alleles[marker],
                                 key=lambda x: -alleles[marker][x]):
                seq = allele if seqformat is None \
                    else ensure_sequence_format(allele, seqformat,
                        library=library, marker=marker)
                reportfile.write("%i\tALLELE\t%s\n" %
                    (alleles[marker][allele], seq))
            seq = top_noise[marker][0] if seqformat is None \
                else ensure_sequence_format(top_noise[marker][0],
                    seqformat, library=library, marker=marker)
            reportfile.write("%i\tNOISE\t%s\n\n" % (top_noise[marker][1], seq))
130
131
132
133
134
            noisy_markers += 1
            alleles[marker] = {}

    # Drop this sample completely if it has too many noisy markers.
    if noisy_markers > max_noisy:
jhoogenboom's avatar
jhoogenboom committed
135
        reportfile.write("Sample %s appears to be contaminated!\n\n" % tag)
136
137
138
139
140
141
        return

    # The sample is OK, write out its alleles.
    for marker in alleles:
        for allele in sorted(alleles[marker],
                             key=lambda x: -alleles[marker][x]):
jhoogenboom's avatar
jhoogenboom committed
142
143
            seq = allele if seqformat is None else ensure_sequence_format(
                allele, seqformat, library=library, marker=marker)
jhoogenboom's avatar
jhoogenboom committed
144
145
            outfile.write("\t".join(
                [tag, marker, str(alleles[marker][allele]), seq]) + "\n")
146
#find_alleles_sample
jhoogenboom's avatar
jhoogenboom committed
147
148


149
150
151
152
153
154
155
156
157
def get_max_expected_alleles(max_alleles, marker, library):
    if max_alleles is not None:
        return max_alleles
    if "max_expected_copies" in library:
        return library["max_expected_copies"].get(marker, 2)
    return 2
#get_max_expected_alleles


jhoogenboom's avatar
jhoogenboom committed
158
def add_arguments(parser):
159
    add_input_output_args(parser, False, False, True)
jhoogenboom's avatar
jhoogenboom committed
160
161
162
    filtergroup = parser.add_argument_group("filtering options")
    filtergroup.add_argument('-m', '--min-allele-pct', metavar="PCT",
        type=float, default=_DEF_MIN_ALLELE_PCT,
jhoogenboom's avatar
jhoogenboom committed
163
        help="call heterozygous if the second allele is at least this "
jhoogenboom's avatar
jhoogenboom committed
164
165
             "percentage of the highest allele of a marker "
             "(default: %(default)s)")
jhoogenboom's avatar
jhoogenboom committed
166
167
    filtergroup.add_argument('-M', '--max-noise-pct', metavar="PCT",
        type=float, default=_DEF_MAX_NOISE_PCT,
168
169
        help="a sample is considered contaminated/unsuitable for a marker if "
             "the highest non-allelic sequence is at least this percentage of "
jhoogenboom's avatar
jhoogenboom committed
170
             "the highest allele of that marker (default: %(default)s)")
jhoogenboom's avatar
jhoogenboom committed
171
172
173
    filtergroup.add_argument('-n', '--min-reads', metavar="N",
        type=pos_int_arg, default=_DEF_MIN_READS,
        help="require at least this number of reads for the highest allele "
jhoogenboom's avatar
jhoogenboom committed
174
             "of each marker (default: %(default)s)")
jhoogenboom's avatar
jhoogenboom committed
175
    filtergroup.add_argument('-a', '--max-alleles', metavar="N",
176
177
178
179
        type=pos_int_arg,
        help="allow no more than this number of alleles per marker; if "
             "unspecified, the amounts given in the library file are used, "
             "which have a default value of 2")
jhoogenboom's avatar
jhoogenboom committed
180
181
    filtergroup.add_argument('-x', '--max-noisy', metavar="N",
        type=pos_int_arg, default=_DEF_MAX_NOISY,
182
183
        help="entirely reject a sample if more than this number of markers "
             "have a high non-allelic sequence (default: %(default)s)")
jhoogenboom's avatar
jhoogenboom committed
184
    filtergroup.add_argument('-c', '--stuttermark-column', metavar="COLNAME",
jhoogenboom's avatar
jhoogenboom committed
185
186
187
        help="name of column with Stuttermark output; if specified, sequences "
             "for which the value in this column does not start with ALLELE "
             "are ignored")
jhoogenboom's avatar
jhoogenboom committed
188
    add_sequence_format_args(parser)
jhoogenboom's avatar
jhoogenboom committed
189
190
191
192
#add_arguments


def run(args):
193
194
    files = get_input_output_files(args)
    if not files:
jhoogenboom's avatar
jhoogenboom committed
195
196
        raise ValueError("please specify an input file, or pipe in the output "
                         "of another program")
197

198
199
200
201
    find_alleles(files[0], files[1], args.report, args.min_reads,
                 args.min_allele_pct, args.max_noise_pct, args.max_alleles,
                 args.max_noisy, args.stuttermark_column, args.sequence_format,
                 args.library)
jhoogenboom's avatar
jhoogenboom committed
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
#run


def main():
    """
    Main entry point.
    """
    parser = argparse.ArgumentParser(
        description=__doc__)
    try:
        add_arguments(parser)
        run(parser.parse_args())
    except OSError as error:
        parser.error(error)
#main


if __name__ == "__main__":
    main()