Detection and characterization of genomics variations in complex diseases

Abstract

Advances in genetic research and sequencing technologies over the past decades have driven the development of many cutting-edge mathematical methods and bioinformatic tools for detecting genetic variations ranging from single point mutations to large-scale structural variations. Most of the tools were proposed for detecting germline variations, while the detection of somatic mutations, especially those with low allele frequencies, remains very challenging. This limitation is an impediment to progress in early diagnosis and prevention of diseases, such as cancer and aging-related diseases, especially since low-frequency somatic mutations are believed to play important roles underlying disease aetiology. Attention deficit hyperactivity disorder (ADHD) and schizophrenia are two complex psychiatric disorders resulting from complex interactions between genetic and environmental risk factors. Previous studies have discovered many disease associated genes, yet the genetic architectures of these two diseases remain incomplete. Accordingly, the first goal of this thesis is to provide a new method named LFMD, for Low-Frequency Mutations Detection, based on high-depth short-read genome sequencing data. LFMD is a likelihood-based model accounting for PCR duplicates from both strands of original DNA to reduce background noise and unveil low frequency mutations. Based on a site-by-site sensitivity evaluation model, LFMD was demonstrated to achieve superior sensitivity and specificity compared with other state-of-the-art methods. Strikingly, the “barcode free” strategy of LFMD makes it by far the lowest-cost tool. The mitochondrial heterogeneity analysis of 28 samples across different stages of Alzheimer’s Disease showed that the oxidative damage related mutation, C:G>A:T, is significantly enriched in the mid-stage group. This result is consistent with the Mitochondrial Free Radical Theory of Aging, suggesting that Alzheimer’s disease may be linked to the aging of brain cells induced by oxidative damage. However, this was not demonstrated by previous low-frequency mutation studies, suggesting that the method with better sensitivity and precision is essential for detecting low-frequency mutations, which is the prerequisite of elucidating their role in disease aetiology. The second goal of this thesis is to explore the genetic risk factors of ADHD and schizophrenia using the state-of-the-art whole genome sequencing (WGS) and PacBio long-read sequencing technologies. The 50 de novo mutations (45 SNVs, 3 Indels, 2 CNVs) identified in ADHD provide evidence that the potassium channel genes (GALNT8, KCNK3, KCNJ3, and KCNQ5) contribute to ADHD aetiology. The candidate genes (KCNJ12, OPRM1, CACNA1B, MAP2K3, KMT2C, SYN3 and NXPE1) discovered in schizophrenia demonstrated that genes involved in neurogenesis, neuron development, and neural signal transmission, are involved in the aetiology of schizophrenia. To sum up, this thesis presents a new method for detecting low-frequency mutations and explored the genetic basis of two complex diseases, ADHD and schizophrenia using state-of-the-art genome sequencing technologies. LFMD is particularly useful in scenarios pursuing high precision, such as in drug resistance prediction, mitochondrial heterogeneity analysis, cancer screening and early diagnosis. The genetic variants and affected genes identified in this thesis extend our current understanding of the genetic risk factors of ADHD and schizophrenia.