|
| 1 | +# Author: Lenna X. Peterson |
| 2 | +# github.com/lennax |
| 3 | +# arklenna at gmail dot com |
| 4 | + |
| 5 | +import logging |
| 6 | +import sys |
| 7 | +import warnings |
| 8 | + |
| 9 | + |
| 10 | +from parser import Reader, Writer |
| 11 | + |
| 12 | + |
| 13 | +class SampleFilter(object): |
| 14 | + """ |
| 15 | + Modifies the vcf Reader to filter each row by sample as it is parsed. |
| 16 | +
|
| 17 | + """ |
| 18 | + |
| 19 | + def __init__(self, infile, outfile=None, filters=None, invert=False): |
| 20 | + # Methods to add to Reader |
| 21 | + def get_filter(self): |
| 22 | + return self._samp_filter |
| 23 | + |
| 24 | + def set_filter(self, filt): |
| 25 | + self._samp_filter = filt |
| 26 | + if filt: |
| 27 | + self.samples = [val for idx, val in enumerate(self.samples) |
| 28 | + if idx not in set(filt)] |
| 29 | + |
| 30 | + def filter_samples(fn): |
| 31 | + """Decorator function to filter sample parameter""" |
| 32 | + def filt(self, samples, *args): |
| 33 | + samples = [val for idx, val in enumerate(samples) |
| 34 | + if idx not in set(self.sample_filter)] |
| 35 | + return fn(self, samples, *args) |
| 36 | + return filt |
| 37 | + |
| 38 | + # Add property to Reader for filter list |
| 39 | + Reader.sample_filter = property(get_filter, set_filter) |
| 40 | + Reader._samp_filter = [] |
| 41 | + # Modify Reader._parse_samples to filter samples |
| 42 | + self._orig_parse_samples = Reader._parse_samples |
| 43 | + Reader._parse_samples = filter_samples(Reader._parse_samples) |
| 44 | + self.parser = Reader(filename=infile) |
| 45 | + # Store initial samples and indices |
| 46 | + self.samples = self.parser.samples |
| 47 | + self.smp_idx = dict([(v, k) for k, v in enumerate(self.samples)]) |
| 48 | + # Properties for filter/writer |
| 49 | + self.outfile = outfile |
| 50 | + self.invert = invert |
| 51 | + self.filters = filters |
| 52 | + if filters is not None: |
| 53 | + self.set_filters() |
| 54 | + self.write() |
| 55 | + |
| 56 | + def __del__(self): |
| 57 | + try: |
| 58 | + self._undo_monkey_patch() |
| 59 | + except AttributeError: |
| 60 | + pass |
| 61 | + |
| 62 | + def set_filters(self, filters=None, invert=False): |
| 63 | + """Convert filters from string to list of indices, set on Reader""" |
| 64 | + if filters is not None: |
| 65 | + self.filters = filters |
| 66 | + if invert: |
| 67 | + self.invert = invert |
| 68 | + filt_l = self.filters.split(",") |
| 69 | + filt_s = set(filt_l) |
| 70 | + if len(filt_s) < len(filt_l): |
| 71 | + warnings.warn("Non-unique filters, ignoring", RuntimeWarning) |
| 72 | + |
| 73 | + def filt2idx(item): |
| 74 | + """Convert filter to valid sample index""" |
| 75 | + try: |
| 76 | + item = int(item) |
| 77 | + except ValueError: |
| 78 | + # not an idx, check if it's a value |
| 79 | + return self.smp_idx.get(item) |
| 80 | + else: |
| 81 | + # is int, check if it's an idx |
| 82 | + if item < len(self.samples): |
| 83 | + return item |
| 84 | + filters = set(filter(lambda x: x is not None, map(filt2idx, filt_s))) |
| 85 | + if len(filters) < len(filt_s): |
| 86 | + # TODO print the filters that were ignored |
| 87 | + warnings.warn("Invalid filters, ignoring", RuntimeWarning) |
| 88 | + |
| 89 | + if self.invert: |
| 90 | + filters = set(xrange(len(self.samples))).difference(filters) |
| 91 | + |
| 92 | + # `sample_filter` setter updates `samples` |
| 93 | + self.parser.sample_filter = filters |
| 94 | + if len(self.parser.samples) == 0: |
| 95 | + warnings.warn("Number of samples to keep is zero", RuntimeWarning) |
| 96 | + logging.info("Keeping these samples: {0}\n".format(self.parser.samples)) |
| 97 | + return self.parser.samples |
| 98 | + |
| 99 | + def write(self, outfile=None): |
| 100 | + if outfile is not None: |
| 101 | + self.outfile = outfile |
| 102 | + if self.outfile is None: |
| 103 | + _out = sys.stdout |
| 104 | + elif hasattr(self.outfile, 'write'): |
| 105 | + _out = self.outfile |
| 106 | + else: |
| 107 | + _out = open(self.outfile, "wb") |
| 108 | + logging.info("Writing to '{0}'\n".format(self.outfile)) |
| 109 | + writer = Writer(_out, self.parser) |
| 110 | + for row in self.parser: |
| 111 | + writer.write_record(row) |
| 112 | + |
| 113 | + def _undo_monkey_patch(self): |
| 114 | + Reader._parse_samples = self._orig_parse_samples |
| 115 | + delattr(Reader, 'sample_filter') |
0 commit comments