assoc¶
Perform genome-wide association study (GWAS) using mixed linear models.
Use this command with GRM input from grm.
Basic Syntax¶
gelex assoc -b genotypes -p phenotypes.tsv --grm my_grm -o gwas_run
gelex assoc --pheno <pheno_file> --bfile <genotype_prefix> --grm <grm_prefix...> [OPTIONS]
Required inputs are phenotype file (--pheno), genotype prefix (--bfile),
and at least one GRM prefix (--grm).
Method Selection¶
Pick model and preprocessing strategy before tuning runtime options.
Warning
If you use --transform with --model d, dominance signals may be
attenuated, which can reduce power to detect dominance effects.
Options¶
Quick Start Options
-p, --phenorequiredPhenotype TSV file in format
FID IID trait1 ....-b, --bfilerequiredPLINK binary prefix (
.bed/.bim/.fam).--grmrequiredOne or more GRM prefixes.
-o, --outgelexOutput prefix for GWAS results.
Input and Covariate Options
--pheno-col20-based trait column index in phenotype file.
--qcovarQuantitative covariate TSV in format
FID IID covar1 ....--dcovarCategorical covariate TSV in format
FID IID factor1 ....
Model Configuration
--modelaAssociation model:
a(additive) ord(dominance).--geno-methodOrthCenterHWEGenotype processing method. Available methods:
StandardizeHWE(SH),CenterHWE(CH),OrthStandardizeHWE(OSH),OrthCenterHWE(OCH),Standardize(S),Center(C),OrthStandardize(OS),OrthCenter(OC). Abbreviations accepted. See Genotype Processing Methods.--transformnonePhenotype transform:
none,dint(Direct INT),iint(Indirect INT).--int-offset0.375INT offset parameter
k.
REML and Performance
--max-iter100Maximum REML iterations.
--tol1e-06REML convergence tolerance.
-c, --chunk-size10000Number of SNPs per association-testing chunk.
-t, --threadshalf of available CPU coresNumber of CPU threads to use.
--locofalseEnable leave-one-chromosome-out analysis.
Output Files¶
After a successful run, GWAS summary statistics are written to:
File pattern |
Contents |
Reference |
|---|---|---|
|
SNP-wise test statistics (effect size, SE, P-value, allele fields) |
Warnings and Notes¶
Warning
--loco requires chromosome-wise GRM inputs generated from
gelex grm --loco. Use the matching GRM prefix in --grm.
Note
For --model d, provide GRM inputs consistent with the dominance model
setup (typically additive + dominance GRMs).
Examples¶
gelex assoc \
-b genotypes_qc \
-p phenotypes.tsv \
--grm my_grm \
-o basic_gwas
gelex assoc \
-b genotypes_qc \
-p phenotypes.tsv \
--grm my_grm \
--qcovar pcs.tsv \
--dcovar sex.tsv \
-o covar_gwas
gelex assoc \
-b genotypes_qc \
-p phenotypes.tsv \
--grm my_grm.add \
--loco \
-o loco_gwas
gelex assoc \
-b genotypes_qc \
-p phenotypes.tsv \
--grm my_grm.add my_grm.dom \
--model d \
--transform iint \
-o dom_gwas
gelex assoc \
-b genotypes_qc \
-p phenotypes.tsv \
--grm my_grm \
--threads 16 \
-o fast_gwas
See Also¶
grm for preparing GRM inputs.
GWAS Results (.gwas.tsv) for GWAS output columns.