A cautionary note on ignoring polygenic background when mapping quantitative trait loci via recombinant congenic...

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Gene mapping often involves testing numerous genetic loci for association with phenotypes.
• Complex genetic traits are frequently modeled using linear mixed models (LMMs) that account for major genes and polygenic background effects.
• Simplifying analyses by omitting the polygenic random effects term from LMMs is common but problematic.

Purpose of the Study:

• To demonstrate the detrimental impact of ignoring random effects on type I error rates in gene mapping.
• To introduce and validate a novel bootstrap procedure for genetic linkage analysis using recombinant congenic strains within an LMM framework.
• To provide a reliable alternative for controlling statistical errors in complex genetic trait analysis.

Main Methods:

• Simulation studies were conducted to assess type I error rates when random effects are omitted from LMMs.
• A new bootstrap procedure was developed for genetic linkage mapping with recombinant congenic strains under LMMs.
• The performance of the proposed bootstrap method was evaluated through simulations, comparing its error rates to nominal levels.

Main Results:

• Ignoring the random effects term in LMMs leads to unacceptably high type I error rates in gene mapping.
• The proposed bootstrap procedure demonstrated type I error rates close to nominal levels in simulations.
• Slight inflation of type I error rates was observed with the bootstrap method when the random effects variance was large.

Conclusions:

• Omitting polygenic random effects in genetic linkage analysis significantly compromises the control of type I errors.
• The novel bootstrap procedure offers a robust solution for accurate statistical inference in gene mapping with recombinant congenic strains.
• Researchers should be cautious about potential modeling issues when simple linear regression yields multiple significant linkage peaks, suggesting the need for LMMs.