Study On QTL Mapping Using High-density SNP Markers Cross The Entire Genome

It has been a major challenge that a large numbers of false positive results produced dueto multiple testing in Genome-wide association study, which is still an important andunsolved problem. The probability for every SNP of being in linkage disequilibrium withQTL evaluated in expectation conditional maximum algorithm was be able to using in QTLmapping. The results from both GWAS and ECM were often the SNPs with significantassociation with traits in statistics, however, it’s still important issue that the way judging theapproximate position of QTL using these SNP markers.In our study, a kind of two-stage combined strategy integrating association analysis andECM algorithm was suggested to locate QTL using the whole genomic SNP markers throughselecting markers and re-estimating for the same population. Two different combinations wererecommended according to the order of use (Combined I and Combined II). Furthermore, wetried two kinds of methods, fixed interval (1cM) and LD analysis (LD I and LD II), todetermine the approximate position of QTL and generated a length of segment, according tothese significant SNP markers resulted from combined strategy.Results of study are：(1) There are a large number of false positive results in GWAS no matter the P values ofhypothesis were adjusted using Bonferroni correction or FDR (False Positive Rate) method.The false positive rate of QTL detection are0.74for Bonferroni and0.77for FDR. Combinedstrategy are able to reducing the false positive rate remarkably, and the false positive rate are0.22for Combined I and0.13for Combined II.13of15major QTL (explaining more than1%of the total genetic variation and being detected in multiple regression analysis) werelocated by both two methods, and several additional QTLs were also detected by Combined I.It’s the fact that association between SNPs were abated through selecting step and theadvantage of ECM were integrated, which confirmed the success rate of QTL mapping andcontrolled the false positive rate and further raised the accurate of QTL analysis.(2) The results of QTL mapping was affected by population size and the heritability ofquantitative trait. The larger population size, the higher success rate and accurate of QTLlocation. An enough number of individuals were needed to acquired satisfied results, and it’sbetter that unrelated individuals were used. The higher h heritability, the higher accurate ofQTL location, but the ability of detection maybe influenced by the effect size of QTL.(3) The approximate position of QTL can be deduced accurately when the LD analysis methods were used, and a length of interval were generated.13major QTLs were accuratelydetected by LD I method and15by LD II. However, the intervals from LD I were morenarrow than intervals from LD II, especially for the situation that the distance from significantSNP to QTL was very close. The false rate generated by LD analysis method was lower thanfixed interval, which raised the accurate of QTL analysis.In sum, the combined strategy and LD analysis method discussed in our study can raise theaccurate of QTL analysis, which is valuable to the gene identification work in the next step.