Novel bioinformatics methods for integrative detection of structural variants from long-read sequencing

点击翻译按钮获取中文摘要

Project Summary/Abstract Structural variants (SVs) are the largest source of variations in the human genome and are frequently associated with disease phenotypes. Thus, the identification and characterization of SVs are essential for understanding human genome structure and function. The goal of this proposal is to develop a generalized SV calling pipeline that can leverage information from the latest developments in sequencing technology and human reference genome representations to discover and resolve SVs at high accuracy. I will first integrate information across sequencing platforms to increase SV calling accuracy. Multiple sequencing and mapping platforms are now used to detect SVs from human genome data. My pipeline will increase the accuracy of SV calling with a data integration model that handles a diverse set of genomic platforms. I will next develop a novel SV scoring model based on genomic context and coverage. Several factors, such as the generally low sequence coverage in typical long-read studies, as well as alignment errors due to highly repetitive sequences, can result in a potentially high rates of false positives for SVs when using parameters for high-sensitivity calling. I use two sets of important features of SVs, genomic context and coverage, into a machine-learning model to compute confidence in SV calls for downstream analysis. Finally, I will add support for graph genome alignments by implementing support for sequence data aligned to graph genome assemblies in GFA file format. Unlike single reference genomes, pangenomes are particularly useful for characterizing large-scale structural differences in genomes between different ethnicity groups. Pangenomes would bring us closer to capturing the full extent of human genomic variation, and thus represent an important resource to leverage for SV calling. In summary, in this project I will develop a generalized SV calling pipeline capable of integrating multiple technical platforms for discovering SVs and providing support for future developments in pangenome graph assemblies. With the research training plan, I will 1) gain expertise in genomics and bioinformatics, 2) promote diversity in biomedical research though involvement in educational efforts in the community, 3) develop oral and written communication skills, and 4) prepare a scientific career focused on the study and education of human genome variation.