Пример #1
0
    def test_filter_n(self):
        ht = khmer.new_hashtable(10, 4**10)

        filename = utils.get_test_data('simple_2.fa')
        outname = utils.get_temp_filename('test_filter.out')

        (total_reads, n_consumed) = ht.consume_fasta(filename)
        assert total_reads == 4, total_reads
        assert n_consumed == 63, n_consumed

        (total_reads, n_seq_kept) = khmer.filter_fasta_file_any(ht, filename,
                                                                total_reads,
                                                                outname, 1)
        assert n_seq_kept == 3, n_seq_kept

        names = load_fa_seq_names(outname)
        assert names == ['1', '2', '3']
Пример #2
0
    def test_filter_n(self):
        ht = khmer.new_hashtable(10, 4**10)

        filename = utils.get_test_data('simple_2.fa')
        outname = utils.get_temp_filename('test_filter.out')

        (total_reads, n_consumed) = ht.consume_fasta(filename)
        assert total_reads == 4, total_reads
        assert n_consumed == 63, n_consumed

        (total_reads,
         n_seq_kept) = khmer.filter_fasta_file_any(ht, filename, total_reads,
                                                   outname, 1)
        assert n_seq_kept == 3, n_seq_kept

        names = load_fa_seq_names(outname)
        assert names == ['1', '2', '3']
Пример #3
0
    def test_filter(self):
        ht = khmer.new_hashtable(10, 4**10)

        filename = os.path.join(thisdir, 'test_data/simple_1.fa')
        outname = os.path.join(self.tempdir, 'test_filter.out')

        (total_reads, n_consumed) = ht.consume_fasta(filename)
        assert total_reads == 3, total_reads
        assert n_consumed == 63, n_consumed

        (total_reads, n_seq_kept) = khmer.filter_fasta_file_any(ht, filename,
                                                                total_reads,
                                                                outname, 2)
        assert n_seq_kept == 2, n_seq_kept

        names = load_fa_seq_names(outname)
        assert names == ['1', '2']