Autozygome sequencing expands the horizon of human knockout research and provides novel insights into human phenotypic variation

Abstract

The use of autozygosity as a mapping tool in the search for autosomal recessive disease genes is well established. We hypothesized that autozygosity not only unmasks the recessiveness of disease causing variants, but can also reveal natural knockouts of genes with less obvious phenotypic consequences. To test this hypothesis, we exome sequenced 77 well phenotyped individuals born to first cousin parents in search of genes that are biallelically inactivated. Using a very conservative estimate, we show that each of these individuals carries biallelic inactivation of 22.8 genes on average. For many of the 169 genes that appear to be biallelically inactivated, available data support involvement in modulating metabolism, immunity, perception, external appearance and other phenotypic aspects, and appear therefore to contribute to human phenotypic variation. Other genes with biallelic inactivation may contribute in yet unknown mechanisms or may be on their way to conversion into pseudogenes due to true recent dispensability. We conclude that sequencing the autozygome is an efficient way to map the contribution of genes to human phenotypic variation that goes beyond the classical definition of disease.

Figure 1. Allele frequency distribution according to mutation type.

Each box represents the 25^th to 75^th percentiles, with the median shown as a line and the mean as a cross. The whiskers represent the 5^th and 95^th percentiles. Non-parametric ANOVA (Kruskal-Wallis test) indicates significant differences in the group medians at p<0.0001 and all pairwise median comparisons were also significant (p<0.0001, Mann-Whitney non parametric test), except for frameshift vs. nonsense.

Figure 2. Longer ROHs are more enriched for homozygous LoFs.

Solid line: Percentage of LoFs found within an ROH of indicated length or longer. Dashed line: Percentage of autosomal genome bases included (on average) in ROHs at the indicated length or longer. Dotted line: Percentage of gain in LoF recovery compared to genomic coverage at the indicated ROH length or longer (the percent ratio of the two curves minus one).

Figure 3. LoF alleles that are significantly biased (at p<0.1 using a one-tailed binomial test, see Text S1 for details) to be within the autozygome (when called at a minimum ROH cutoff of 2MB) are rarer compared to alleles biased to be outside of the autozygome.

The boxes and whiskers represent allele frequency distributions as in Fig. 1. The two groups have a significantly different median at p<0.0005 (Mann-Whitney non parametric test).