Skip to content

mecfs_bio.assets.reference_data.polyfun.annotations.baseline_lf_annotations

Reference assets: the allele-bearing baseline-LF 2.2.UKB annotation parquets and the single sorted annotation matrix built from them.

The annotations come from the ~30GB baselineLF_v2.2.UKB.polyfun.tar.gz bundle, which is streamed and never stored: only the per-chromosome .annot.parquet members (~1GB total, carrying A1/A2) are kept. The derived ~19.5M x 187 matrix and the members directory are path_remap candidates (large, few-file, rarely read) for relocation to /mnt/d.

Attributes:

BASELINE_LF_ANNOTATION_MATRIX module-attribute 

BASELINE_LF_ANNOTATION_MATRIX = BuildBaselineLFAnnotationParquetTask.create(
    asset_id="baseline_lf_2.2_ukb_annotations",
    annot_members_task=BASELINE_LF_ANNOTATION_PARQUET_MEMBERS,
)

BASELINE_LF_ANNOTATION_PARQUET_MEMBERS module-attribute 

BASELINE_LF_ANNOTATION_PARQUET_MEMBERS = StreamExtractAnnotationParquetsTask(
    meta=ReferenceDataDirectoryMeta(
        group="polyfun",
        sub_group="annotations",
        sub_folder=PurePath("raw"),
        id=AssetId(
            "baseline_lf_2.2_ukb_annot_parquet_members"
        ),
    ),
    url="https://broad-alkesgroup-ukbb-ld.s3.amazonaws.com/UKBB_LD/baselineLF_v2.2.UKB.polyfun.tar.gz",
)