Megalencephaly syndromes: exome pipeline strategies for detecting low-level mosaic mutations

Abstract

Two megalencephaly (MEG) syndromes, megalencephaly-capillary malformation (MCAP) and megalencephaly-polymicrogyriapolydactyly-hydrocephalus (MPPH), have recently been defined on the basis of physical and neuroimaging features. Subsequently, exome sequencing of ten MEG cases identified de-novo postzygotic mutations in PIK3CA which cause MCAP and de-novo mutations in AKT and PIK3R2 which cause MPPH. Here we present findings from exome sequencing three unrelated megalencephaly patients which identified a causal PIK3CA mutation in two cases and a causal PIK3R2 mutation in the third case. However, our patient with the PIK3R2 mutation which is considered to cause MPPH has a marked bifrontal band heterotopia which is a feature of MCAP. Furthermore, one of our patients with a PIK3CA mutation lacks syndactyly/polydactyly which is a characteristic of MCAP. These findings suggest that the overlap between MCAP and MPPH may be greater than the available studies suggest. In addition, the PIK3CA mutation in one of our patients could not be detected using standard exome analysis because the mutation was observed at a low frequency consistent with somatic mosaicism. We have therefore investigated several alternative methods of exome analysis and demonstrate that alteration of the initial allele frequency spectrum (AFS), used as a prior for variant calling in samtools, had the greatest power to detect variants with low mutant allele frequencies in our 3 MEG exomes and in simulated data. We therefore recommend non-default settings of the AFS in combination with stringent quality control when searching for causal mutation(s) that could have low levels of mutant reads due to post-zygotic mutation.

Figure 1. Variants identified in sample 2 by the full and flat AFS models including the mosaic post zygotic variant in PIK3CA.

Figure 2. Evaluation of select methods of exome analysis using simulated data.

(A) Minimum mutant allele frequencies for identification of a simulated variant at total read depths ranging from 4 to 100. (B) Base call accuracies of a simulated variant identified at the minimum mutant allele frequency. As total depth increases, the minimum percentage of the alternative or mutant allele that is required to identify the variant decreases until a total depth is reached at which an additional read with the alternate allele is required to detect the variant. At these total read depths, the addition of an alternate read causes the mutant allele frequency and the base call accuracy to increase. Consequently, the plots of minimum alternate allele percentage (A) and corresponding base call accuracies (B) are composed of a series of spikes at which points an additional alternate read was required to identify the variant.