The Development Of QTL Mapping Software Package GAS_NC?

Many important traits in plants are quantitative traits.By locating associated favorable genes and predicting hybrid performance can elucidate the genetic basis,improve crop genetic structure and boost the efficiency for breeding.Genetic analysis for quantitative trait in plants is aimed at parental segregation populations like F2,DH and BC.However,these groups only have two parents,in order to explore elite alleles,it is necessary to develop genetic analysis research of multiple parents breeding population such as NCII,NCIII and diallel crossing.There is little research done on QTL mapping of these groups,especially the epistatic QTL mapping.For the convenience of application work,this research devloped software that has graphic user interface named GAS_NCII.It contains epistatic QTL mapping methods in NCII,devloped by our own research group.This software package has three models:Empirical Bayes,Empirical Bayes lasso and Mass number of effects.The main results are as follows:1.The Empirical Bayes module is to use phenotypic genotypic data of parents and hybrid F1 as data input to perform epistatic QTL mapping.Genotypic data uses individual as rows and SNP marker as columns.When F1's data is not given,then one has to input information data of mating design which corresponding the Hybrid F1 design option under data input menu in software.This model can handle Main effect module,Epistatic model and Environmental interaction model(Main + Epistatic + Environment-by-Main + Environment-by-Epistatic),one can use tree view in software to choose module.The parameter estimation methods of this model are Empirical Bayes and Segregation analysis + Empirical Bayes.When data is input and module choosen,click the run button to run the program.QTL are display as numbers,and one can input Marker name file to change it to gene name.The result contains QTL type,Marker,Chi-value,Effect,LOD,P-value and r^2.we use this software to analysis the genetic foundation of oil content.As a result,8 main-effect QTL and 37 interacted QTL were found,this is the same as SAS software package.2.The difference between EBL AS SO and Empirical Bayes modules is in the method of parameter estimation.This module has a Cross-Validation method.If you don't do a Cross-Validation,you can choose a=0.01 and b=0.01,which has big shrinkage.If one does Cross-Validation,he can input one number into Fold number in Cross-Validation experiment to generate the best values for parameters a and b,which control shrinkage degree.The module of this model can include hundreds of thousands of effects.We used this module in analysis of two years data of cotton diallel cross group and we found 9 epistatic QTL in quantitative traits.These results are the same as those calculated by R package.3.Mass number of effects estimation module is used to estimate effect more than 20 thousands,it can be used to handle Epistatic module.The parameter estimation method contains Empirical Bayes and Empirical Bayes lasso with Corss-Validation,the input and output are the same as those in the Empirical Bayes module.This research implement a Monte Carlo simulation for 440 individual and 700 SNPs.The main and epistatic effects have a total number of 980000 parameters to be estimated.Simulation locus contains additive,dominant,additive-by-additive,additive-by-dominance,dominance-by-additive and dominant-by-dominant.Simulation was repeated 100 times.The results showed that at heritability of 10%,the detection power of 2 main-effect QTL were 100%and 99%respectively,the detection power of 4 epistatic QTL were 97%,94%,96%and 85%respectively,flase positive rate was 0.0131 ‰,absolute deviation of estimated QTL effect is less than 0.18,the standard deviation of absolute deviation is less than 0.27.