Abstract
The Quantitative Genetic Analysis Station (QGAStation) is a software package that has been developed to perform statistical analysis for complex traits. It consists of five domains for handling data from diallel crosses, regional trials, core germplasm collections, QTL mapping, and microarray experiments. The first domain contains genetic models for diallel cross analysis, in which genetic variance components and genetic-by-environment interactions can be estimated, and genetic effects can be predicted. The second domain evaluates the performance of varieties in regional trials by implementing a general statistical method that outperforms ANOVA in tackling unbalanced data that arises frequently in trials across multiple locations and over a number of years. The third domain, using predicted genotypic values as proxy, constructs core germplasm collections covering sufficient genetic diversity with lower redundancy. The fourth domain manages genotypic and phenotypic data for QTL mapping. Linkage maps can be constructed and genetic distances can be estimated; the statistical methods that have been implemented apply to both chiasmatic and achiasmatic organisms. Another part of this domain can filter systematic noises in phenotypic data. The fifth domain focuses on the cDNA expression data that is generated by microarray experiments. A two-step strategy has been implemented to detect differentially expressed genes and to estimate their effects. Except in the fourth domain, the major statistical methods that have been used are mixed linear model approaches that have been implemented in the C language. Computational efficiency is further boosted for computers that are equipped with graphics processing units (GPUs). A user friendly graphic interface is provided for Microsoft Windows and Apple Mac operating systems. QGAStation is available at http://ibi.zju.edu.cn/software/qga/ .
Highlights
The Quantitative Genetic Analysis Station (QGAStation) is a software package that has been developed to perform statistical analysis for complex traits
In the mixed linear models, best linear unbiased prediction (BLUP) is used to predict random effects if the variance components are estimated with restricted maximum likelihood (REML), whereas linear unbiased prediction (LUP) or adjusted unbiased prediction (AUP) is the options in QGAStation if the variance components are estimated by minimum norm quadratic unbiased estimation (MINQUE) [11]
In quantitative genetics, when studying the genetic factors that underlie a trait of interest, laboratory work and statistical analyses are naturally complementary
Summary
The statistical models implemented in QGAStation are mainly in the form of mixed linear models that are useful in handling complicated genetic models even for unbalanced data. In the mixed linear models, best linear unbiased prediction (BLUP) is used to predict random effects if the variance components are estimated with REML, whereas linear unbiased prediction (LUP) or adjusted unbiased prediction (AUP) is the options in QGAStation if the variance components are estimated by MINQUE [11]. A set of genetic experimental data would contain observations of a single trait, with one observation per individual. QGAStation implements a general conditioning method to take care of both types of datasets [13,14]. A conditioning method is embedded into the diallel cross analysis and the QTL mapping domains in QGAStation
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
