Statistical guidance for experimental design and data analysis of mutation detection in rare monogenic mendelian diseases by exome sequencing

Abstract

Recently, whole-genome sequencing, especially exome sequencing, has successfully led to the identification of causal mutations for rare monogenic Mendelian diseases. However, it is unclear whether this approach can be generalized and effectively applied to other Mendelian diseases with high locus heterogeneity. Moreover, the current exome sequencing approach has limitations such as false positive and false negative rates of mutation detection due to sequencing errors and other artifacts, but the impact of these limitations on experimental design has not been systematically analyzed. To address these questions, we present a statistical modeling framework to calculate the power, the probability of identifying truly disease-causing genes, under various inheritance models and experimental conditions, providing guidance for both proper experimental design and data analysis. Based on our model, we found that the exome sequencing approach is well-powered for mutation detection in recessive, but not dominant, Mendelian diseases with high locus heterogeneity. A disease gene responsible for as low as 5% of the disease population can be readily identified by sequencing just 200 unrelated patients. Based on these results, for identifying rare Mendelian disease genes, we propose that a viable approach is to combine, sequence, and analyze patients with the same disease together, leveraging the statistical framework presented in this work.

Conflict of interest statement

Competing Interests: The authors have declared that no competing interests exist.

Figures

Figure 1. The calculated power of exome sequencing for rare monogenic Mendelian diseases for various parameter combinations.

Figure 2. Genes underlying highly heterogeneous diseases can be identified by sequencing a moderate sized sample.

The calculated power with varying degrees of genetic heterogeneities (R) ranging from 0.01 to 1 is shown. Upper panel: power of Tr for detecting a recessive gene; Middle panel: power difference Tr-Ta for detecting a recessive gene; Lower panel: power of Td for detecting a dominant gene. Other parameters are fixed to the default values: number of mutations m = 300; total number of genes M = 20,000; sensitivity of detecting mutations Ps = 0.8; and the mutation probability equals genome-wide average w = 1. See Tables S2, S3 and S4 for more dense sampling of R values. Note that power does not always increase monotonously with sample sizes (zigzag line patterns). The loss of power upon increase of sample size is related to discrete changes in the significance level cutoff of the test and thus very small test size (not close to 0.05) as shown in Table S1, since the distribution of the test statistic is discrete.

Figure 3. High sensitivity of detecting mutations is required to achieve a useful power.

The power for varying degrees of sensitivities of mutation detection, ranging from 0.1 to 1 is shown. Other parameters are fixed to the default values: number of mutations m = 300; total number of genes M = 20,000; genetic heterogeneity R = 0.05; and the mutation probability equals genome-wide average w = 1. See Tables S5 and S6 for more dense coverage of sensitivities of mutation detection.

Figure 4. Strict filtering of false positives has limited impact on recessive diseases but dramatically reduces the power of detecting dominant disease genes.

The power for varying degrees of filtering efficiencies, ranging from 5 to 500, is shown. Upper panel: power of Tr for recessive data; Lower panel: power of Td for dominant data. Other parameters are fixed to the default values: genetic heterogeneity R = 0.05; total number of genes M = 20,000; sensitivity of detecting mutations Ps = 0.8; and the mutation probability equals genome-wide average w = 1. See Tables S7 and S8 for more dense coverage of filtering efficiencies.

Figure 5. Power is low for long genes.

The power for varying degrees of relative mutation probabilities, ranging from 0.1 to 10 times the genome average is shown. Upper panel: power of Tr for recessive data; Lower panel: power of Tr for recessive data. Other parameters are fixed to the default values: number of mutations m = 300; genetic heterogeneity R = 0.05; total number of genes M = 20,000; and sensitivity of detecting mutations Ps = 0.8. See Tables S9 and S10 for more dense coverage of filtering efficiencies.