Bayesian model for accurate MARSALA (mutated allele revealed by sequencing with aneuploidy and linkage analyses)

This study is aimed at increasing the accuracy of preimplantation genetic test for monogenic defects (PGT-M).

We applied Bayesian statistics to optimize data analyses of the mutated allele revealed by sequencing with aneuploidy and linkage analyses (MARSALA) method for PGT-M. In doing so, we developed a Bayesian algorithm for linkage analyses incorporating PCR SNV detection with genome sequencing around the known mutation sites in order to determine quantitatively the probabilities of having the disease-carrying alleles from parents with monogenic diseases. Both recombination events and sequencing errors were taken into account in calculating the probability.

Data of 28 in vitro fertilized embryos from three couples were retrieved from two published research articles by Yan et al. (Proc Natl Acad Sci. 112:15964-9, 2015) and Wilton et al. (Hum Reprod. 24:1221-8, 2009). We found the embryos deemed "normal" and selected for transfer in the previous publications were actually different in error probability of 10^-4-4%. Notably, our Bayesian model reduced the error probability to 10^-6-10^-4%. Furthermore, a proband sample is no longer required by our new method, given a minimum of four embryos or sperm cells.

The error probability of PGT-M can be significantly reduced by using the Bayesian statistics approach, increasing the accuracy of selecting healthy embryos for transfer with or without a proband sample.