Related Experiment Videos
Modifying the Schwarz Bayesian information criterion to locate multiple interacting quantitative trait loci
Malgorzata Bogdan1, Jayanta K Ghosh, R W Doerge
1Institute of Mathematics, Wroclaw University of Technology, 50-370 Wroclaw, Poland.
Genetics
|July 9, 2004
Summary
A modified Schwarz Bayesian information criterion (BIC) improves the detection of quantitative trait loci (QTL) interactions. This new method accurately identifies main effects and pairwise interactions, overcoming limitations of standard criteria.
Area of Science:
- Genetics
- Statistical Genetics
- Bioinformatics
Background:
- Locating multiple interacting quantitative trait loci (QTL) is often modeled as a multiple regression problem.
- Estimating the number of interacting QTL and their interactions presents a significant challenge in statistical genetics.
- Existing model selection criteria are inadequate for estimating interaction terms, leading to overestimation of QTL numbers.
Purpose of the Study:
- To address the overestimation of QTL numbers caused by epistatic terms in standard model selection criteria.
- To investigate the behavior of the Schwarz Bayesian information criterion (BIC) in the context of QTL analysis.
- To propose a novel modification of BIC for improved detection of main effects and pairwise interactions.
Main Methods:
- Framing QTL localization as a multiple regression problem using marker genotypes as regressors.
- Analyzing the tendency of standard model selection criteria to overestimate interactions when main effects are absent.
- Developing a modified BIC to specifically detect main effects and pairwise interactions in genetic data.
- Conducting extensive simulations to evaluate the performance of the modified BIC in backcross populations.
Main Results:
- Standard model selection criteria tend to overestimate the number of interactions and QTL when epistatic terms are present without main effects.
- The proposed modified BIC effectively detects both main effects and pairwise interactions.
- Simulation studies confirm that the modified BIC performs well in practice for identifying interacting QTL.
Conclusions:
- The modified BIC offers a robust solution for accurately estimating the number of interacting QTL, including main effects and pairwise interactions.
- This methodology enhances the precision of QTL mapping by mitigating overestimation issues.
- The approach is extensible to diverse populations and higher-order genetic interactions.