Variants detection in regions of segmental duplication facilitates molecular diagnosis for IEI such as chronic granulomatous disease caused by NCF1 mutations

Abstract

With the expanding application of next-generation sequencing (NGS) in clinical practice, rapid progress has been made in molecular diagnoses for diseases such as inborn errors of immunity (IEI). However, due to characteristic short read length and fragment size for NGS, genomic regions with segmental duplication pose a big challenge for accurate mutation detection. Reads originating from segmental duplications are mapped with high ambiguity and are often discarded by most variant callers, leading to poor variant identification. Dozens of known IEI causal genes are located in regions of segmental duplications, such as NCF1, IGLL1, FCGR3A, etc. It is important to increase the sensitivity of mutation detections in such regions.In this study, we develop a strategy to significantly increase the sensitivity of identifying rare germline variants in regions of segmental duplication for IEI patients while controlling the number of false positives. By forcing multi-aligned reads from segmental duplications to map to functionally significant coding regions in the region concerned, all possible small variants are being identified. This is followed by the identification of false positives originating from intrinsic sequence differences among genomic duplications and the identification of potential common variants by matching with an in-house database.Using simulated data and real-world data from the 'Genome In a Bottle’ project, we demonstrate a significant improvement in variant detection sensitivity, ranging from 80% to 98% and 75% to 94%, respectively. This is accompanied by a limited increase in false positives that can be dealt with by downstream analysis. Applying our method to whole exome sequencing data led to the identification of the common ΔGT mutation in NCF1 in three patients with the chronic granulomatous disease, demonstrating the power of this novel approach.In summary, we developed SDrecall, a bioinformatic method that is by far the most robust open-source approach to detect rare, disease-causing variants in regions of segmental duplication. The method should apply to both IEI and other Mendelian diseases with causal mutations located in such regions and is suited to WGS, WES, and targeted sequencing data.