Parallel in PLINK for linear association for SNP effects

Question

[landscape95]

Hi, I am doing linear association calculations for SNP effects, I see it take a lot of time to do, does anyone have any experience in parallelizing this job. I have used --threads <num of cores> but it says it only uses 1 thread

here is my code for the calculation in plink:

plink --bfile ukbb_dis --pheno phenotype_dis.pheno --mpheno 4 --allow-no-sex --covar phenotype_dis.pheno --covar-number 5-21 --missing-phenotype -10000000 --linear --adjust --out assoc_SNP_height_linear --threads 4

Here is the current output:

257682 MB RAM detected; reserving 128841 MB for main workspace.
1133273 variants loaded from .bim file.
284516 people (0 males, 0 females, 284516 ambiguous) loaded from .fam.
Ambiguous sex IDs written to assoc_SNP_height_linear.nosex .
284516 phenotype values present after --pheno.
Using 1 thread.
Warning: This run includes BLAS/LAPACK linear algebra operations which
currently disregard the --threads limit.  If this is problematic, you may want
to recompile against single-threaded BLAS/LAPACK.
--covar: 17 out of 21 covariates loaded.
Before main variant filters, 284516 founders and 0 nonfounders present.
Calculating allele frequencies... done.
Total genotyping rate is 0.992737.
1133273 variants and 284516 people pass filters and QC.
Phenotype data is quantitative.
Writing linear model association results to
assoc_SNP_height_linear.assoc.linear ... 0%

Your help is really appreciated!

Answer 1

[chrchang523]

It’s plink 2.0, not 1.9, that has all the major optimizations for —linear. Using v2.0 for this should speed up the job by >100x.

[landscape95]

Thank you very much Chang, I'm really glad to have the author to help me out. Would you please have a look in another topic here: Plink error: Skipping --glm regression on phenotype 'PHENO1' since variance inflation factor for covariate 'COVAR1' is too high. I have some problem now, I think it's not really related to the current topic so I opened a new topic. Hope you would be happy to have a look. Thank you. Best.