A computational framework for predicting the impact of mutations in autism

描述(由申请人提供)：精神障碍患者家庭的下一代基因组测序正在识别数千种对疾病有不同影响的突变 风险。基因社区在将他们的发现转化为疾病机制之前面临的两个主要问题是：(1)如何区分致病突变和中性突变？(2)每个致病突变破坏了什么生物学功能(S)而导致疾病？这些都是需要紧急关注的根本性问题。已发现的大量突变和受影响基因的功能异质性不允许开发一种通用的实验性高通量方法来解决这些问题。尽管现代技术(如CRISPR)为这一方向的未来提供了希望，但在我们能够将其应用于数千种突变之前，还有很长的路要走。因此，需要预测性的计算方法来帮助确定基因和突变的优先顺序和功能特征。在这里，我们建议开发这样的方法，并将它们应用于在自闭症谱系障碍(ASD)家族中识别的编码和非编码变体。我们专注于ASD的遗传变异影响模型将把不同种类的遗传数据与大脑特定的功能数据源整合在一起，例如基因表达和大脑剪接异构体相互作用网络，这些网络是专门为大脑过程量身定做的。我们的方法的独特之处在于，它从预测携带基因突变(S)编码的蛋白质的生物学功能开始，然后进行基因和变体的优先排序和功能影响分配，最后以ASD早期诊断的风险模型结束。我们将通过以下具体目标来实现这些目标：(1)预测基因的生物学功能和孤独症特定候选基因的优先顺序；(2)预测编码和非编码变异对ASD高危因素的功能影响；(3)通过表征突变对蛋白质相互作用、转录和翻译的影响来实验验证预测；(4)开发从外显子或基因组序列估计ASD风险的算法。这是ASD研究中的一种新方法，它结合了分子精神病学、分子生物学和计算生物学不同领域的专业知识。调查人员及其合作者的广泛专业知识确保了对这一问题采取有原则和全面的方法。

DESCRIPTION (provided by applicant): Next-generation genome sequencing of families afflicted by mental disorders is identifying thousands of mutations with variable effect on disease risk. Two main problems that genetic community is facing before it could move forward with translating their findings into disease mechanisms are: (1) how to distinguish the pathogenic disease-causative mutations from the neutral ones? (2) what biological function(s) does each pathogenic mutation disrupt to cause a disease? These are fundamental questions that need urgent attention. The large number of identified mutations and functional heterogeneity of the affected genes do not permit developing a generalized experimental high-throughput method for addressing these problems. Although modern technologies (such as CRISPR) provide hope for the future in this direction, there is still a long way before we are able to apply it to thousandsof mutations. Thus, predictive computational approaches that aid in genes and mutations prioritization and functional characterization are needed. Here, we propose to develop such methods and apply them to coding and non-coding variants identified in families with Autism Spectrum Disorders (ASD). Our ASD-focused model of genetic variant impact will integrate heterogeneous genetic data with brain-specific functional data sources, such as gene expression and brain splice isoform interaction networks that are uniquely tailored towards brain processes. The unique feature of our approach is that it starts with prediction of biological function of a protein encoded by a gene carrying mutation(s), proceeds with gene and variant prioritization and functional impact assignment, and ends with a risk model for early ASD diagnosis. We will accomplish these goals through the following specific aims: (1) Predicting Biological Function of Genes and Autism-Specific Candidate Gene Prioritization; (2) Predicting Functional Impact of Coding and Non-coding Variants Conferring High Risk for ASD; (3) Experimental Validation of Predictions by Characterizing the Effect of Mutations on Protein Interactions, Transcription and Translation; (4) Developing Algorithms for Estimating ASD Risk from the Exome or Genome Sequence. This is as a novel approach in ASD research that combines expertise from diverse areas of molecular psychiatry, molecular biology and computational biology. The broad range of expertise by the investigators and their collaborators ensures a principled and comprehensive approach to the problem.