def test_parse_metadata_df(self):
mini_gctoo = mini_gctoo_for_testing.make()
# convert row_metadata to np.nan
mini_row_meta = mini_gctoo.row_metadata_df.replace([-666, "-666", -666.0], [np.nan, np.nan, np.nan])
logger.debug("mini_row_meta.shape: {}".format(mini_row_meta.shape))
logger.debug("mini_row_meta.index: {}".format(mini_row_meta.index))
logger.debug("mini_row_meta.columns: {}".format(mini_row_meta.columns))
logger.debug("mini_row_meta.dtypes: {}".format(mini_row_meta.dtypes))
gctx_file = h5py.File("functional_tests/mini_gctoo_for_testing.gctx", "r")
row_dset = gctx_file[row_meta_group_node]
col_dset = gctx_file[col_meta_group_node]
# with convert_neg_666
row_df = parse_gctx.parse_metadata_df("row", row_dset, True)
logger.debug("row_df.dtypes: {}".format(row_df.dtypes))
pandas_testing.assert_frame_equal(mini_row_meta, row_df)
# no convert_neg_666
mini_gctoo_with_neg_666 = mini_gctoo_for_testing.make(convert_neg_666=False)
col_df = parse_gctx.parse_metadata_df("col", col_dset, False)
pandas_testing.assert_frame_equal(mini_gctoo_with_neg_666.col_metadata_df, col_df)
# test that ID's are not converted to numeric
expected_rids = [str(i) for i in range(3)]
row_dset = {"id": MockHdf5Dset(expected_rids, str),
"other_meta": MockHdf5Dset(range(3, 6), str)}
r = parse_gctx.parse_metadata_df("row", row_dset, True)
logger.debug("test that ID's are not converted to numeric - r: {}".format(r))
logger.debug("r.index: {}".format(r.index))
self.assertEqual(set(expected_rids), set(r.index))
def test_parse_metadata_df(self):
mini_gctoo = mini_gctoo_for_testing.make()
# convert row_metadata to np.nan
mini_row_meta = mini_gctoo.row_metadata_df.replace([-666, "-666", -666.0], [np.nan, np.nan, np.nan])
gctx_file = h5py.File("functional_tests/mini_gctoo_for_testing.gctx", "r")
row_dset = gctx_file[row_meta_group_node]
col_dset = gctx_file[col_meta_group_node]
# with convert_neg_666
row_df = parse_gctx.parse_metadata_df("row", row_dset, True)
assert_frame_equal(mini_row_meta, row_df)
# no convert_neg_666
mini_gctoo_with_neg_666 = mini_gctoo_for_testing.make(convert_neg_666=False)
col_df = parse_gctx.parse_metadata_df("col", col_dset, False)
assert_frame_equal(mini_gctoo_with_neg_666.col_metadata_df, col_df)