Local ancestry-specific polygenic score#

To calculate PGS for each local ancestry background:

# the local ancestry should be called in advance and saved to <plink2_prefix>.lanc
# see https://kangchenghou.github.io/admix-kit/prepare-dataset.html#step-2-local-ancestry-inference
# to properly format local ancestry file
# apply --dset-build only when the plink file are not in the same build as weight
admix calc-partial-pgs \
    --plink-path <plink2_prefix>.pgen \
    --weights-path <weight_tsv_path> \
    --dset-build 'hg38->hg19' \
    --out out

To also calculate reference PGS for reference populations:

admix calc-partial-pgs \
    --plink-path <plink2_prefix>.pgen \
    --weights-path <weight_tsv_path> \
    --ref-plink-path 1kg-plink \
    --ref-pops 'CEU,YRI' \
    --dset-build 'hg38->hg19' \
    --out out

# to download the 1,000 Genomes reference plink files, use the following command:
admix get-1kg-ref --dir=1kg-ref --build=hg38

Parameter options#

admix.cli.calc_partial_pgs(plink_path: str, weights_path: str, out: str, ref_plink_path: str | None = None, ref_pops: List[str] | None = None, weight_col: str = 'WEIGHT', ref_pop_col: str = 'Population', dset_build: str | None = None, weights_build: str | None = None)[source]#

Calculate PGS from a weight file and a PLINK file.

Parameters:

  • plink_path (str) –

    Path to plink files. Some examples are:

    • /path/to/chr21.pgen

    • /path/to/genotype/directory

    • /path/to/file_list.txt containing rows of file names

  • weights_path (str) – path to PGS weights, containing CHROM, SNP, REF, ALT, WEIGHT columns

  • out (str) – prefix of the output files. {out}.sample_pgs.tsv and {out}.ref_pgs.tsv will be written to disk.

  • ref_plink_path (str) – path to reference plink files. The ref_plink_path should be a single plink2 file.

  • ref_pops (list of str) – list of populations in reference plink files. For example, “CEU,YRI”

  • weight_col (str, optional) – column in ‘weights_path’ representing the weight, by default “WEIGHT”

  • dset_build (str, optional) – genome build transform for the dataset (“hg19->hg38” or “hg38->hg19”), by default None

  • weights_build (str, optional) – genome build transform for the weights (“hg19->hg38” or “hg38->hg19”), by default None