Additional file 1 of Breed identification using breed-informative SNPs and machine learning based on whole genome sequence data and SNP chip data

Additional file 1:Table S1. Summary of the machine learning classification models. Table S2. Accuracies of different pipelines for breed identification with different most breed-informative SNPs. Table S3. Accuracies of difference machine learning methods with respect to different reference population size and different numbers of most breed-informativeSNPs revealed by DFI. Table S4. Accuracies of difference machine learning methods with respect to different number of breeds. The reference population size was 30 individuals per breed and the most breed-informativeSNPs were from DFI. Table S5. Accuracies when both reference and test population were genotyped with SNP chipsand with sequence. The reference population size was 30individuals per breed and the most breed-informativeSNPs were from DFI. Table S6. The GBCof the misclassified Brown Swiss animals estimated using 789K SNPs. Table S7. Distance matrixamong the 13 cattle breeds. Table S8. Pathway enrichment of genes in the vicinity of the 1,000 most breed-informative SNPs.