Skip to content

Convert h5ad to scdl

main()

Parse the arguments to process the single cell collection.

Source code in bionemo/scdl/scripts/convert_h5ad_to_scdl.py 
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
def main():
    """Parse the arguments to process the single cell collection."""
    parser = argparse.ArgumentParser()
    parser.add_argument(
        "--num-workers", type=int, default=4, help="The number of AnnData loaders to run in parallel [4]."
    )
    parser.add_argument(
        "--use-mp",
        action="store_true",
        default=False,
        help="Use a subprocess for each worker rather than a lightweight OS thread [False].",
    )
    parser.add_argument(
        "--data-path",
        type=str,
        required=True,
        help="A path containing AnnData files. Note: These will all be concatenated.",
    )
    parser.add_argument(
        "--save-path", required=True, type=str, help="An output path where an SCDataset will be stored."
    )

    parser.add_argument(
        "--data-dtype",
        type=str,
        default=None,
        help="The data type to use for the SCDataset. Must be one of 'uint8', 'uint16', 'uint32', 'uint64', 'float32', 'float64'.",
    )
    parser.add_argument(
        "--paginated-load-cutoff",
        type=int,
        default=10_000,
        help="The cutoff in MB for paginated loading of the AnnData files.",
    )
    parser.add_argument(
        "--load-block-row-size",
        type=int,
        default=1_000_000,
        help="The number of rows to load into memory at a time for paginated loading of the AnnData files.",
    )
    args = parser.parse_args()

    with tempfile.TemporaryDirectory() as temp_dir:
        coll = SingleCellCollection(temp_dir)
        coll.load_h5ad_multi(
            args.data_path,
            max_workers=args.num_workers,
            use_processes=args.use_mp,
            data_dtype=args.data_dtype,
            paginated_load_cutoff=args.paginated_load_cutoff,
            load_block_row_size=args.load_block_row_size,
        )
        coll.flatten(args.save_path, destroy_on_copy=True)