Covariates and sample alignment¤

Covariate and offset tables are tab-delimited and must contain exactly one case-insensitive iid or #iid column. Covariate tables require at least one data column. The reader recognizes NA, an empty field, NULL, NaN, and nan as missing values. An optional fid column is ignored. jaxQTL aligns every input on the intersection of sample IDs and preserves the genotype source order.

IID values must use the same spelling in every input. Samples missing from any required input are excluded by the intersection, so compare the IDs before mapping and confirm the resulting sample count.

iid       age  sex  genotype_pc1
sample_1  42   F    -0.013
sample_2  57   M     0.021

Use --covar-name to retain selected covariates or --rm-covar to remove selected covariates. The options are mutually exclusive. --one-hot encodes string columns, and --normalize-covar centers and scales numeric covariates.

Common choices include ancestry PCs, age, sex, and study or processing batches. Choose covariates for the study design rather than treating this list as required. If you use expression PCs in a cell-type-specific analysis, compute them separately for each cell type.

jaxQTL adds an intercept by default. Use --no-intercept only when the design matrix already contains the intended mean structure.

Every input must identify a sample at most once. Duplicate IIDs make alignment ambiguous and cause setup to fail before mapping begins.

Sample inclusion files supplied with --keep or --exclude contain one IID per line and are applied to genotype samples before the cross-input intersection.

Identifier lists¤

Gene (--gene-list), keep-sample (--keep), and exclude-sample (--exclude) files contain one identifier per line. A first line beginning with # is treated as a header and skipped. Sample lists use IIDs; gene lists use phenotype IDs from the phenotype matrix.