OBJECTIVETo explore the feasibility of applying autosomal single nucleotide polymorphisms (SNPs) on parentage testing.

METHODSAll SNP genotyping results of HapMap (r27) were downloaded from the website. With self-made computer programs, SNPs were extracted when their minor allele frequency (MAF) were ≥ 0.30 among all of the 11 HapMap populations. Ninety-six SNPs were chosen and integrated into the Illumina Goldengate bead arrays on the condition that no linkage disequilibrium was found between them. Three father-child-mother trios (9 samples in total) were tested with the arrays. Cumulative paternity index (CPI) was then calculated and compared with genotyping results using 15 short tandem repeats (STRs)(Identifiler(TM)).

RESULTSFamily 1 was found to have nine SNPs or seven STRs that did not conform to the Mendelian laws, Family 2 had 13 such SNPs or seven STRs, and Family 3 only had one such SNP but no STR. For Family 3, when all of the 96 SNPs were used in combine, the CPI was 1207, which had contrasted with the CPI by the 15 STRs, i.e., 355 869.

CONCLUSIONWhen applied to paternity testing, the paternity exclusion (PE) value for a SNP is usually less than 1/3 of that of a STR. The proportion of SNPs not comforming to the Mendelian laws for the tested SNPs may not be as high as that of inconsistent STRs over all tested STRs. Because of the low mutation rate of a SNP, the CPI will be greatly reduced even if one SNP did not conform to the Mendelian laws. Therefore, highly accurate testing methods are required to reduce artificial errors when applying SNPs for paternity testing.