ABI Innovation: Creating Complete and Accurate Alternative Splicing Repertoires from RNA-seq Data

Johns Hopkins University, Baltimore MD

Investigators

Abstract

The project aims to produce a suite of bioinformatics tools to accurately and comprehensively identify alternative splicing variation and present this information to researchers through a web accessible database. Alternative splicing is an intrinsic property of eukaryotic genes. A gene can be spliced in multiple ways to produce different variants depending on the tissue, developmental stage, or disease versus normal condition, with each variant having a distinct function. Creating a repertoire of alternative splicing variations of genes is therefore critical for understanding the biology of a species. Recently, fast and cost-effective next generation sequencing has made it possible to explore the repertoire of genes and their splicing variations in a wide variety of species. However, interpreting the short sequencing reads into gene and alternative splicing annotations has significant challenges, and existing tools cannot always reconstruct alternative splicing patterns with high accuracy. The project seeks to produce, first, a bioinformatics tool to identify alternative splicing variations from sequencing reads in more detail and more accurately than previously possible. Secondly, the methods will be applied to build a comprehensive catalog of alternative splicing events in several plant species, taking advantage of the growing amounts of data being generated. The project will produce open source software that can be used by biologists for both research and education, and will contribute to the larger effort to recruit students from diverse backgrounds into interdisciplinary science, by creating research opportunities for undergraduate and high-school students through summer internships. Alternative splicing (AS) is a widespread mechanism with an important role in creating gene and functional diversity. Next generation sequencing of cellular RNA (RNA-seq) has made it possible to explore the repertoire of splice variants in a cell type or species. However, current bioinformatics tools have difficulty in capturing AS variation in detail, and have not scaled up with the increasing number of samples in a typical experiment. To answer these needs, the first aim is to develop a first-of-its-kind next-generation tool for reconstructing gene and splice variants from short RNA-seq reads, simultaneously from large numbers of samples. It will capture AS variation both more fully and more accurately than current programs, including fine variations and hard-to-detect types of variations. The approach is to combine linear programming techniques, with statistical models of RNA-seq "noise", and with scalable splice graph representations of genes, and exploits similarity between expression profiles of related samples to increase efficiency. The second aim is to produce the first comprehensive catalog of AS events in plant species, with web-based visualizations and analysis capabilities, building on the splice graph model as a compact representation of genes. All software will be open source and available without restrictions for all, for research and education. Additionally, the project will create new teaching materials and opportunities for training and recruiting the next generation of scientists, in particular from among under-represented groups and Baltimore inner-city high-schools students, through Summer internships. All software, teaching materials, and the database will be accessible from http://ccb.jhu.edu/people/florea/research/ .