bghomstats.py 7.8 KB
Newer Older
jhoogenboom's avatar
jhoogenboom committed
1
2
#!/usr/bin/env python
"""
jhoogenboom's avatar
jhoogenboom committed
3
4
5
6
7
Compute allele-centric statistics for background noise in homozygous
reference samples (min, max, mean, sample variance).

Compute a profile of recurring background noise for each unique allele
in the database of reference samples.  The profiles obtained can be used
jhoogenboom's avatar
jhoogenboom committed
8
9
10
11
by bgcorrect to filter background noise from samples.  If many reference
samples are heterozygous (as is usually the case with forensic STR
markers), it is preferable to use bgestimate instead, since it can
handle heterozygous samples as well.
jhoogenboom's avatar
jhoogenboom committed
12
"""
13
from ..lib import pos_int_arg, add_input_output_args, get_input_output_files,\
14
                  add_allele_detection_args, parse_allelelist,\
15
                  get_sample_data, add_sequence_format_args, adjust_stats,\
jhoogenboom's avatar
jhoogenboom committed
16
                  add_random_subsampling_args
jhoogenboom's avatar
jhoogenboom committed
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43

__version__ = "0.1dev"


# Default values for parameters are specified below.

# Default minimum amount of background to consider, as a percentage of
# the highest allele.
# This value can be overridden by the -m command line option.
_DEF_THRESHOLD_PCT = 0.5

# Default minimum number of reads to consider.
# This value can be overridden by the -n command line option.
_DEF_THRESHOLD_ABS = 5

# Default minimum number of samples for each true allele.
# This value can be overridden by the -s command line option.
_DEF_MIN_SAMPLES = 2

# Default minimum number of samples required for each background product
# to be included in the analysis, as a percentage of the number of
# samples with a certain true allele.
# This value can be overridden by the -S command line option.
_DEF_MIN_SAMPLE_PCT = 80.


def add_sample_data(data, sample_data, sample_alleles, min_pct, min_abs):
jhoogenboom's avatar
jhoogenboom committed
44
45
46
47
48
49
50
51
52
53
    # Check presence of all alleles.
    for marker in sample_alleles:
        allele = sample_alleles[marker]
        if (marker, allele) not in sample_data:
            raise ValueError(
                "Missing allele %s of marker %s!" % (allele, marker))
        elif 0 in sample_data[marker, allele]:
            raise ValueError(
                "Allele %s of marker %s has 0 reads!" % (allele, marker))

jhoogenboom's avatar
jhoogenboom committed
54
55
56
57
58
59
    # Enter the read counts into data and check the thresholds.
    for marker, sequence in sample_data:
        if marker not in sample_alleles:
            # Sample does not participate in this marker.
            continue
        allele = sample_alleles[marker]
jhoogenboom's avatar
jhoogenboom committed
60
        factors = [100./x for x in sample_data[marker, allele]]
jhoogenboom's avatar
jhoogenboom committed
61
62
63
64
65
66
67
68
        if (marker, allele) not in data:
            data[marker, allele] = {}
        if sequence not in data[marker, allele]:
            data[marker, allele][sequence] = [None, None, 0]
        for direction in (0, 1):
            data[marker, allele][sequence][direction] = adjust_stats(
                sample_data[marker, sequence][direction] * factors[direction],
                data[marker, allele][sequence][direction])
jhoogenboom's avatar
jhoogenboom committed
69
70
71
        if sum(count >= min_abs and count*factor >= min_pct
               for count, factor in
               zip(sample_data[marker, sequence], factors)):
jhoogenboom's avatar
jhoogenboom committed
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
            data[marker, allele][sequence][2] += 1
#add_sample_data


def filter_data(data, min_samples, min_sample_pct):
    """
    Remove all alleles from data that have less than min_samples samples
    and remove all stats of sequences that don't pass the detection
    thresholds in at least min_sample_pct per cent of the samples with a
    particular allele.  Also add explicit zeros to the stats of the
    sequences that were not seen in all samples with a given allele.
    """
    for marker, allele in data.keys():
        if data[marker, allele][allele][2] < min_samples:
            del data[marker, allele]
            continue
        factor = 100./data[marker, allele][allele][2]
        for sequence in data[marker, allele].keys():
            if data[marker, allele][sequence][2] * factor < min_sample_pct:
                del data[marker, allele][sequence]
                continue
            for i in range(data[marker, allele][sequence][0]["n"],
                           data[marker, allele][allele][2]):
                for direction in (0, 1):
                    adjust_stats(0, data[marker, allele][sequence][direction])
#filter_data


100
101
102
def compute_stats(samples_in, outfile, allelefile, annotation_column, min_pct,
                  min_abs, min_samples, min_sample_pct, seqformat, library,
                  marker, limit_reads, drop_samples):
jhoogenboom's avatar
jhoogenboom committed
103

104
    # Parse allele list.
jhoogenboom's avatar
jhoogenboom committed
105
106
107
108
109
    allelelist = {} if allelefile is None \
                    else parse_allelelist(allelefile, seqformat, library)

    # Read sample data.
    data = {}
110
    get_sample_data(
111
        samples_in,
112
113
114
115
116
117
        lambda tag, sample_data: add_sample_data(
            data, sample_data,
            {m: allelelist[tag][m].pop() for m in allelelist[tag]},
            min_pct, min_abs),
        allelelist, annotation_column, seqformat, library, marker, True,
        limit_reads, drop_samples)
jhoogenboom's avatar
jhoogenboom committed
118
119
120
121
122

    # Ensure minimum number of samples per allele and filter
    # insignificant background products.
    filter_data(data, min_samples, min_sample_pct)

jhoogenboom's avatar
jhoogenboom committed
123
124
125
    outfile.write("\t".join(["marker", "allele", "sequence", "n", "fmin",
                     "fmax", "fmean", "fvariance", "rmin", "rmax", "rmean",
                     "rvariance"]) + "\n")
jhoogenboom's avatar
jhoogenboom committed
126
127
    for marker, allele in data:
        for sequence in data[marker, allele]:
jhoogenboom's avatar
jhoogenboom committed
128
            outfile.write("\t".join([marker, allele, sequence] + [
jhoogenboom's avatar
jhoogenboom committed
129
                str(x) if abs(x) > 0.0000000001 else "0" for x in (
jhoogenboom's avatar
jhoogenboom committed
130
131
132
133
134
135
136
137
                    data[marker, allele][sequence][0]["n"],
                    data[marker, allele][sequence][0]["min"],
                    data[marker, allele][sequence][0]["max"],
                    data[marker, allele][sequence][0]["mean"],
                    data[marker, allele][sequence][0]["variance"],
                    data[marker, allele][sequence][1]["min"],
                    data[marker, allele][sequence][1]["max"],
                    data[marker, allele][sequence][1]["mean"],
jhoogenboom's avatar
jhoogenboom committed
138
                    data[marker, allele][sequence][1]["variance"])]) + "\n")
jhoogenboom's avatar
jhoogenboom committed
139
140
141
142
#compute_stats


def add_arguments(parser):
143
    add_input_output_args(parser)
jhoogenboom's avatar
jhoogenboom committed
144
    add_allele_detection_args(parser)
jhoogenboom's avatar
jhoogenboom committed
145
146
    filtergroup = parser.add_argument_group("filtering options")
    filtergroup.add_argument('-m', '--min-pct', metavar="PCT", type=float,
jhoogenboom's avatar
jhoogenboom committed
147
148
149
        default=_DEF_THRESHOLD_PCT,
        help="minimum amount of background to consider, as a percentage "
             "of the highest allele (default: %4.2f)" % _DEF_THRESHOLD_PCT)
jhoogenboom's avatar
jhoogenboom committed
150
    filtergroup.add_argument('-n', '--min-abs', metavar="N", type=pos_int_arg,
jhoogenboom's avatar
jhoogenboom committed
151
152
153
        default=_DEF_THRESHOLD_ABS,
        help="minimum amount of background to consider, as an absolute "
             "number of reads (default: %(default)s)")
jhoogenboom's avatar
jhoogenboom committed
154
155
    filtergroup.add_argument('-s', '--min-samples', metavar="N",
        type=pos_int_arg,
jhoogenboom's avatar
jhoogenboom committed
156
157
158
        default=_DEF_MIN_SAMPLES,
        help="require this minimum number of samples for each true allele "
             "(default: %(default)s)")
jhoogenboom's avatar
jhoogenboom committed
159
160
    filtergroup.add_argument('-S', '--min-sample-pct', metavar="PCT",
        type=float,
jhoogenboom's avatar
jhoogenboom committed
161
162
163
164
        default=_DEF_MIN_SAMPLE_PCT,
        help="require this minimum number of samples for each background "
             "product, as a percentage of the number of samples with a "
             "particular true allele (default: %(default)s)")
jhoogenboom's avatar
jhoogenboom committed
165
    filtergroup.add_argument('-M', '--marker', metavar="MARKER",
jhoogenboom's avatar
jhoogenboom committed
166
        help="work only on MARKER")
jhoogenboom's avatar
jhoogenboom committed
167
168
    add_sequence_format_args(parser)
    add_random_subsampling_args(parser)
jhoogenboom's avatar
jhoogenboom committed
169
170
171
172
#add_arguments


def run(args):
173
174
    files = get_input_output_files(args)
    if not files:
jhoogenboom's avatar
jhoogenboom committed
175
176
        raise ValueError("please specify an input file, or pipe in the output "
                         "of another program")
177
    compute_stats(files[0], files[1], args.allelelist, args.annotation_column,
jhoogenboom's avatar
jhoogenboom committed
178
179
180
                  args.min_pct, args.min_abs, args.min_samples,
                  args.min_sample_pct, args.sequence_format, args.library,
                  args.marker, args.limit_reads, args.drop_samples)
jhoogenboom's avatar
jhoogenboom committed
181
#run