不同因素对猪INDEL基因型填充准确性的影响研究OA

[Objective]This study investigated genotype imputation for porcine insertions/deletions to evaluate the effects of different factors on imputation accuracy and to identify an optimal imputation strategy.[Method]Based on whole-genome sequencing data from 1119 pigs,two reference panels were constructed:reference panel Ⅰ containing only INDEL genotypes and reference panel Ⅱ containing both SNP and INDEL genotypes.Two hundred Large white pigs were selected as the validation population.Autosomal INDEL genotypes were masked chromosome by chromosome under a completely random scheme at five missing proportions(20％,45％,70％,95％,and 99％)to simulate different marker densities.Five reference panel sizes(10,50,100,500,and 1000)and seven minor allele frequency(MAF)intervals([0.01,0.03),[0.03,0.05),[0.05,0.1),[0.1,0.2),[0.2,0.3),[0.3,0.4),and[0.4,0.5])were set.In addition,using 100 Large white pigs as the core population,four reference panel diversity levels(L0-L3)were established by introducing other pig breeds.The INDEL imputation accuracy of Beagle 5.5,IMPUTE 5,and Minimac 3 was compared using concordance rate(CR)and pearson correlation coefficient(PC).[Result]Under all tested conditions,reference panel Ⅰ showed higher imputation accuracy than reference panel Ⅱ,with mean PC values of 0.797 and 0.760,respectively.In reference panel Ⅰ,the PC values of Beagle 5.5 and IMPUTE 5 decreased from 0.898 and 0.900 at 20％missingness to 0.641 and 0.640 at 99％missingness,respectively,whereas the PC value of Minimac 3 was only 0.481 at 99％missingness.The PC value of Beagle 5.5 increased from 0.619 at a reference panel size of 10 to 0.794 at 100,but increased only to 0.832 at 1000.When MAF increased from 0.01～0.03 to 0.4～0.5,the PC value of Beagle 5.5 increased from 0.571 to 0.837.The PC value at L0 was 0.863,which was higher than those at L1,L2,and L3(0.847,0.847,and 0.848,respectively).[Conclusion]In summary,the inclusion of SNPs in the reference panel reduced the accuracy of INDEL imputation.As marker density decreased,imputation accuracy also declined.Beagle 5.5 and IMPUTE 5 were both suitable for porcine INDEL imputation,whereas Minimac 3 performed poorly under extremely low marker density.Imputation accuracy increased with reference panel size,but the gain in accuracy from additional samples became markedly weaker once the reference panel size reached about 100 individuals.The L0 level,which was genetically most consistent with the validation population,showed the highest accuracy.For porcine INDEL genotype imputation,it is recommended to use data with relatively high marker density,and a reference panel size of about 100 individuals is suggested.An MAF threshold of>0.05 is recommended as a quality-control criterion for imputed INDEL genotypes.When the reference panel consists of multiple breeds,populations with genetic backgrounds similar to that of the validation population should be selected whenever possible.Choosing appropriate software while considering both computational efficiency and imputation accuracy remains critical for ensuring reliable results.This study provides a reference for optimizing porcine INDEL genotype imputation strategies and offers a methodological basis for the genetic dissection of complex traits and breeding applications based on INDEL variation.