Related Experiment Videos
Statistical power of QTL mapping methods applied to bacteria counts
P Tilquin1, W Coppieters, J M Elsen
1Unité de Génétique, Faculté d'ingénierie biologique, agronomique et environnementale, Université catholique de Louvain, Croix du Sud 2 bte 14, B-1348 Louvain-la-Neuve, Belgium.
Genetical Research
|February 28, 2002
Summary
Most quantitative trait loci (QTL) mapping methods assume normal distributions, but bacteria counts are often skewed. Non-parametric (NP) mapping is robust, but parametric methods with mathematical transformations (MT) outperform NP for highly zero-inflated data.
Area of Science:
- Genetics
- Statistical Genetics
- Bioinformatics
Background:
- Quantitative trait loci (QTL) mapping typically assumes normally distributed phenotypes.
- Many biological and agricultural phenotypes, such as bacteria counts (colony-forming units, CFU), exhibit non-normal distributions, often right-skewed with many zero values (ties).
Purpose of the Study:
- To evaluate the efficiency of four QTL mapping methods (least-squares, maximum-likelihood, non-parametric, and nested ANOVA) when applied to non-normally distributed bacteria count data.
- To compare the performance of these methods with and without mathematical transformations (MT) on the data.
Main Methods:
- Simulated bacteria count data with varying proportions of zeros.
- Applied four QTL mapping methods: least-squares (LS), maximum-likelihood (ML), non-parametric (NP), and nested ANOVA (AN).
- Utilized a quantile-based mathematical transformation (MT) to address data skewness and ties.
Main Results:
- Parametric methods (LS, ML, AN) showed a significant loss of power when analyzing raw, non-normally distributed bacteria counts compared to normally distributed data.
- Non-parametric (NP) mapping maintained power without data transformation.
- Mathematical transformation (MT) restored the power of parametric methods to levels comparable to NP mapping.
- Parametric methods combined with MT outperformed NP mapping when the bacteria count data had a very high proportion of zeros (e.g., 70.8%).
Conclusions:
- The asymmetry and ties in phenotypic distributions significantly impact QTL mapping power, particularly for parametric methods.
- Mathematical transformation is crucial for improving the efficiency of parametric QTL mapping methods with skewed count data.
- The choice of QTL mapping method should consider the presence of ties and the distribution of phenotypic data, especially for count-based traits in bacterial disease studies.