| Bulked segregant analysis assisted by next generation sequencing,abbreviated as BSA-seq,has proven to be a simple,efficient and cheap approach for fast mapping of single genes as well as quantitative trait loci(QTL).In recent years,more and more studies of QTL mapping using BSA-seq have been reported,which is gradually becoming a popular approach for QTL mapping.In mapping of a single gene with BSA-seq,the target gene can be directly mapped without the need of significance test in general.For QTL mapping with BSA-seq,however,significance test is necessary because the number of QTLs is not known and great errors exist in the experiment.In general,a genomic region is likely to harbor a QTL only if it exhibits significant allelic difference between the two DNA pools.A complete QTL mapping method should be able to estimate the QTL position and its confidence interval,as well as the QTL effect and heritability(the proportion of phenotypic variation explained by the QTL).In BSA-seq,as the information obtained from the experiment is very limited,estimation of QTL effect and heritability is quite difficult,and no related methods have been reported so far.Up to now,all the statistical methods for BSA-Seq proposed are focused on the point estimation of QTL positions,and the way for estimating confidence intervals is still not well established.In this dissertation,a new statistical framework for QTL mapping based on BSA-Seq is proposed,named block regression mapping(BReM).BReM provides a complete solution for QTL mapping based BSA-Seq,enabling simultaneous estimation of QTL position,confidence interval,effect direction,and heritability.According to the model and calculation method,some factors influencing BSA-Seq(including heritability,selection rate,allele frequency difference,population type,population size,segregation bias etc.)were analyzed.In general,the inferences obtained from the heritability estimation model are consistent with QTL mapping experience.Analysis on these factors can provide theoretical reference for experimental design.To examine the feasibility and efficiency of BReM,the method was used to analyze two sets of published BSA-Seq experimental data(from yeast and rice respectively).In the yeast experiment,the trait of high ethanol tolerance(HET)was investigated,a haploid population was used,and an unsymmetrical pool design was adopted,namely,using a selected pool with extreme phenotype(ethanol tolerance)and a random unselected pool.Four QTLs were mapped by BReM,named as qHET7,qHET13,qHET15a and qHET15b,respectively.Their confidence interval widths of two levels were on average 46.5 kb and 53.75 kb,respectively.The heritability of each QTL varied between 18.41%and 30.93%.The resistant alleles of three QTLs were from the more tolerant parent Seg5,which explained 62.08%of phenotypic variation in total;only of one QTL was from the less tolerant parent BY710,which explained 30.93%of phenotypic variation.The candidate genes of the four QTLs were examined.The most possible candidates were TFC4,COG8,IRA2 and SHE4.Among them,IRA2 has been reported to be related to ethanol tolerance,involved in the metabolism process under ethanol stress.In the rice experiment,the cold tolerance at the seedling stage(CTSS)was investigated using an F3 population of 10,800 individuals.A symmetrical pool design was adopted,namely,a tolerant DNA pool and a sensitive DNA pool were constructed.Six QTLs were mapped by BReM,named as qCTSS1,qCTSS2a,qCTSS2b,qCTSS5,qCTSS8 and qCTSS10,respectively.Their confidence interval widths of two levels were on average 2.09 Mb and 2.82 Mb,respectively.The resistant alleles of four QTLs were from the tolerant parent Nipponbare,which explained 18.93%of phenotypic variation altogether;while of the other two QTLs were from the senstive parent LPBG,which explained 9.69%of phenotypic variation in total.These two practical examples have demonstrated the feasibility and efficiency of BReM. |
PDF Full Text Request