Documentation Analysis Methods QTL and GWAS Mapping
Browse Documentation
Molecular Data Processing QTL and GWAS Mapping Integrated Trait–Gene Analysis Methods

Analysis Methods

QTL and GWAS Mapping

Understand molecular association tests, genomic windows and the meaning of reported thresholds

Molecular QTL mapping

MatrixEQTL tests genotype associations with normalized expression, splicing or APA features, producing eQTL, sQTL or apaQTL results respectively. These names identify the molecular phenotype of the association; they are not names for TWAS or EWAS models.

Cis and trans
The template defines cis associations on the same chromosome within 200 kb of the molecular feature interval; other tested pairs are trans. A local plot around a gene is a viewing window and does not redefine the association's cis/trans class.
Significance
The supplied mapping scripts use P ≤ 0.05 / N variants, where N is the number retained after mapping-stage genotype QC. The denominator is not the number of variant–feature pairs, and this is not an FDR threshold. Confirm the stored cutoff for the selected dataset.
Effect and sample count
Beta describes change in the transformed molecular phenotype per counted allele. The template defaults to ALT dosage for expression and REF dosage for splicing/APA; compare signs only after aligning allele definitions. Splicing and APA models use observed samples for each feature, so effective sample counts can differ.

GWAS and trait-associated loci

The supplied GWAS workflow uses a GEMMA linear mixed model with genotype principal components and a relatedness matrix, and reports the Wald-test P value. Its significance cutoff is 0.05 divided by the number of variant result rows in that analysis. PlantQTLdb also indexes published GWAS evidence; source studies can use different designs and thresholds.

Variant, locus and gene
Several associated variants can tag one underlying locus through linkage disequilibrium. A variant overlapping a gene or lying nearby supplies positional evidence; it does not establish that the gene mediates the trait.
Counts
Association rows, distinct variants, molecular features and genes are different counting units. Multiple splicing or APA features can map to one gene. Compare counts only when the unit, dataset and selection match.
No qualifying result
No significant result means that no tested association passed the selected rule. Not imported, not tested, QC excluded and not converged describe different availability states and must not be interpreted as a negative biological finding.

Variant consequence annotation

The SnpEff workflow predicts variant consequences using a database built for the matching reference and gene annotation. It restores reference chromosome names and checks reference alleles before annotation, then restores analysis chromosome names while retaining variant IDs.

Read the plots in context

Check the species, reference, dataset, molecular layer and trait before comparing panels. A Manhattan plot places associations by genomic position, whereas the EWAS overview orders molecular features by P value. Thresholds belong to the full analysis; zooming, filtering or plotting fewer background points does not recalculate significance.

Workflow source code

Reviewed code version: f2bde42 (8 November 2026). The workflows expose parameters for new runs; imported result thresholds should be read from the selected analysis.