BLUP (blup)
June 6, 2026 ยท View on GitHub
Use graphld blup to compute infinitesimal-model effect-size weights.
Basic Usage
uv run graphld blup \
/path/to/sumstats.sumstats \
output.tsv \
--heritability 0.1 \
--num-samples 100000
Required Option
| Option | Description |
|---|---|
-H, --heritability | Heritability for the analyzed variant scope, between 0 and 1 |
Inputs And Output
- Input summary statistics:
.sumstats,.vcf/.vcf.gz, or.parquet; see Summary Statistics - Output: a tab-separated file of BLUP weights at the path you pass as
OUT
BLUP also uses the shared LDGM options documented in the CLI overview.
Notes
- BLUP assumes an infinitesimal effect covariance matrix
D; GraphLD setstrace(D)equal to--heritabilityacross the matched LDGM effect indices in the analyzed scope. If you restrict BLUP with--chromosomeor--population, pass the heritability for that restricted scope. - If
--num-samplesis omitted, GraphLD infers sample size from the input file. - Use
--population,--chromosome, and--run-in-serialto limit the scope of a run or simplify debugging.