Research On Genotype Imputation With Whole-genome Sequence Data

Whole-genome sequence(WGS)data plays an important role in animal genetics and breeding.Using WGS datacan improve both the powerof GWAS and the accuracy of genomicselection(GS).However,sequencing thousands of individuals of interest are still too expensive to achieve.Genotype imputation is a good way to solve this problem.In this study,a broiler chicken population were used as experimental population,we genotyped 450 chickens with 600 K SNP array,and utilized G matrixto selected 24 key individuals for whole genome re-sequencing by maximizing the expected genetic relationship.We systematically investigate the impacts of imputation algorithms,key individual selection strategies,reference population size,marker density of target pa nel,sequencing depthof reference panel,and different sequencing strategywith a given total cost of genotypingon imputation accuracywhen imputing array data to sequence data.Finally,these resultsin our study wereshown as follows:1.More marker density of target panel would resulted in more accuracy in genotype imputation,and two-step imputation could improve the accuracy of genotype imputation.2.With fixed sequence depth,the imputation accuracy would increasing as the number of sequenced animals increasing,but the growth slowdown as the number of sequenced animals more than 12.3.With fixed the size of reference,increasing the sequencing depth of reference panel would resulted in more accuracy in genotypeimputation,but the growt h slowdown as thesequence depth of sequenced animals more than 6X.4.Using selected key individuals as reference population for re-sequencing would resulted in a higher imputation accuracy.5.With a given total cost of genotyping,accuracy increased with the size of reference population for FImpute,but the pattern was not valid for Beagle,which showed the highest accuracy at six folds coverage for scenarios used in this study.6.The imputation accuracy of FImpute was better than Beagle in low MAF SNPs。In conclusion,at a fixed sequencing cost,the optimal imputation strategy should take sequencing depth and size of reference population,imputation algorithms,and marker density and population structure of target population into consideration comprehensively and improved imputation accuracy by selected key individuals to re-sequencing.In thisbroiler chicken population,the optimal imputation strategy of FImpute should sequencedmore key animals with low depth;but for Beagle,the optimal imputation strategy should sequenced more key animals at six folds coverage.This work sheds additional light on how to design and execute of genotype imputation in livestock population.