def test_sum_bases_in_bins(self): """Test sum_bases_in_bins""" scg_tsv = ospj(DATA_PATH, "sample0_gt500_scg.tsv") b = sum_bases_in_bins(pd.read_csv(scg_tsv, sep="\t"), ospj(DATA_PATH, "sample0_gt500.fa")) assert_equal(12, b) df = get_approved_bins(ospj(DATA_PATH, "sample0_gt500_scg.tsv"), max_missing_scg=2, max_multicopy_scg=4) b = sum_bases_in_bins(df, ospj(DATA_PATH, "sample0_gt500.fa")) assert_equal(4, b)
def test_write_approved_bins(self): """Test write_approved_bins""" df = get_approved_bins(ospj(DATA_PATH, "sample0_gt500_scg.tsv"), max_missing_scg=2, max_multicopy_scg=4) assert_equal(2, int(df.Cluster)) write_approved_bins(df, ospj(DATA_PATH, "sample0_gt500.fa"), TMP_BASENAME_DIR, "sample0_gt500") ok_(os.path.exists(ospj(TMP_BASENAME_DIR, "sample0_gt500_bin2.fa"))) # make sure both have equal amount of records assert_equal( open(ospj(TMP_BASENAME_DIR, "sample0_gt500_bin2.fa")).read().count(">"), open(ospj(DATA_PATH, "sample0_gt500_bin2.fa")).read().count(">"))
def test_get_approved_bins(self): """Test get_approved_bins""" df = get_approved_bins(ospj(DATA_PATH, "sample0_gt500_scg.tsv"), max_missing_scg=2, max_multicopy_scg=4) assert_equal(2, int(df.Cluster))