Ballgown: A General Statistical Framework for Transcript Assemblies

• Activity Number: 440
• Type: Topic Contributed
• Date/Time: Wednesday, August 6, 2014 : 8:30 AM to 10:20 AM
• Sponsor: ENAR
• Author(s): Alyssa Frazee*+ and Steven Salzberg and Geo Pertea and Jeff Leek
• Companies: and Johns Hopkins University School of Medicine and Johns Hopkins University School of Medicine and Johns Hopkins Bloomberg School of Public Health
• Keywords: RNA-seq ; differential expression ; software ; transcriptome assembly

Abstract

Computational biologists have developed fast, deterministic algorithms that use RNA ­sequencing (RNA-seq) data to perform data-driven transcriptome assembly and to measure each transcript's abundance in a cell population. However, rigorously testing the transcripts for differential expression between populations remains challenging. We have developed a flexible statistical framework, Ballgown, for analyzing variability in a transcriptome assembly and for detecting differential expression at the transcript, exon, or gene level. This framework can be used in conjunction with any assembly tool that constructs transcripts and estimates their associated abundances. As compared to the widely­-used differential expression package Cuffdiff 2, Ballgown is more accurate for two-­class differential expression testing. It is also more flexible: Ballgown handles a wider variety of experimental designs and is capable of adjusting for confounding variables. Finally, Ballgown is computationally more efficient and better-suited for large datasets than Cuffdiff 2. The Ballgown R package is freely available for download from GitHub (https://github.com/alyssafrazee/ballgown).