Detection and phasing of single base de novo mutations in biopsies from human in vitro fertilized embryos by advanced whole-genome sequencing

Abstract

Currently, the methods available for preimplantation genetic diagnosis (PGD) of in vitro fertilized (IVF) embryos do not detect de novo single-nucleotide and short indel mutations, which have been shown to cause a large fraction of genetic diseases. Detection of all these types of mutations requires whole-genome sequencing (WGS). In this study, advanced massively parallel WGS was performed on three 5- to 10-cell biopsies from two blastocyst-stage embryos. Both parents and paternal grandparents were also analyzed to allow for accurate measurements of false-positive and false-negative error rates. Overall, >95% of each genome was called. In the embryos, experimentally derived haplotypes and barcoded read data were used to detect and phase up to 82% of de novo single base mutations with a false-positive rate of about one error per Gb, resulting in fewer than 10 such errors per embryo. This represents a ∼ 100-fold lower error rate than previously published from 10 cells, and it is the first demonstration that advanced WGS can be used to accurately identify these de novo mutations in spite of the thousands of false-positive errors introduced by the extensive DNA amplification required for deep sequencing. Using haplotype information, we also demonstrate how small de novo deletions could be detected. These results suggest that phased WGS using barcoded DNA could be used in the future as part of the PGD process to maximize comprehensiveness in detecting disease-causing mutations and to reduce the incidence of genetic diseases.

Figure 1.

Characteristics of de novo SNVs in embryo #1. After filtering, 110 putative de novo SNVs (including about 10 errors) were identified in embryo #1. (A) Phasing enabled the parent of origin to be determined for 93 of the de novo SNVs. Similar to previous studies, almost twice as many de novo SNVs came from the father as compared to the mother. (B) Specific nucleotide changes for de novo (dark blue) and inherited (light blue) were plotted by frequency. Frequencies of nucleotide changes were similar between de novo and inherited, as would be expected for true de novo SNVs. 95% confidence interval error bars were computed using a one sample proportions test, allowing for Yates’ continuity correction using R software (R Core Team 2014). The error bars suggest that the small differences observed between de novo and inherited are insignificant.

Figure 2.

Detection of heterozygous deletions of small exons. LFR haplotype information can be used to separate coverage for each allele. Normalized coverage from each LFR haplotype for embryo #1 biopsies 1 and 2 and embryo #2, as well as 50-bp read coverage windows for both parents, were plotted (blue indicates father; red, mother). (A) A heterozygous deletion of ∼500 bp in the gene TTC23L removing all of exon 4 and part of the intron on either side in both biopsies of embryo #1 and the father was detected. (B) A heterozygous deletion of ∼1000 bp in the gene SPINK14 removing all of exon 3 and parts of the intron on either side was identified in all three biopsies. Coverage for the parents is more difficult to interpret in this region, but it appears that again the father has less coverage.