Noninvasive prenatal paternity testing using targeted massively parallel sequencing

Recent advances in massively parallel sequencing (MPS) technology have provided efficient methods for noninvasive prenatal paternity testing (NIPAT). However, a well-accepted protocol has not been established. The present study developed an MPS-based approach for NIPAT and compared the performance of two recently reported methods for MPS data interpretation.

We selected 1795 unlinked polymorphic single-nucleotide polymorphisms (SNPs) and performed paternity analysis in 34 real parentage test cases with maternal plasma samples using the Illumina HiSeq platform. Sequencing data were interpreted by the straightforward counting method for the identification of paternal alleles and mathematical algorithms for paternity index (PI) calculation, respectively.

Based on the sequencing data from each family case, both of the two statistical approaches produced a significant separation between the biological father and 90 unrelated males (p < 0.0001) when sufficient effective loci were attained. Nevertheless, up to 30.82% of real paternal alleles were filtered by a predefined cutoff and resulted in insufficient effective loci, especially in plasma samples with low fetal fraction (approx. 90.60% were filtered). In contrast, the PI calculation model utilized all maternal homozygous SNPs as effective loci (approx. 40% of total SNPs) and successfully identified the correct biological father, with the log-transformed combined PI (Lg(CPI)) value varying from 68.23 to 158.01 in each family case.

Our study illustrates that the Bayesian approach represents the better choice in NIPAT data interpretation. Further, the adoption of more informative markers (e.g., tri-allelic SNPs, tetra-allelic SNPs, and micro-haplotypes) or deeper sequencing is recommended for the improvement of the testing efficiency.