Fast annotation-agnostic differential expression analysis...

Fast annotation-agnostic differential expression analysisLeonardo Collado-Torres1, 2, Andrew E. Jaffe2, Jeffrey T. Leek1

1Department of Biostatistics, The Johns Hopkins University Bloomberg School of Public Health,2Lieber Institute for Brain Development, Maltz Research Laboratories

IntroductionSince the development of high-throughput technologies forprobing the genome we have been interested in finding dif-ferences across groups that could potentially explain thephenotypic differences we observe. In other words, meth-ods for generation of hypothesis at a large scale wherewe try our best to remove artifacts. The traditional toolshave focused on the transcriptome and are highly depen-dent on existing annotation. Frazee et al1 developed a sta-tistical framework to find candidate Differentially ExpressedRegions (DERs) without relying on annotation. We have im-plemented a modified version of this approach that is fasterin order to handle larger data sets and whose total process-ing time is comparable to other tools such as DESeq (An-ders et al, Genome Biology 2010).

DERfinder

ResultsConclusionsGoal accomplished: from BAM files to annotated candidateDERs in less than a day!

Comparable time versus other methods (DESeq, ...)

Open questions/todo:

● Reduce memory requirements.● When to merge regions?● How to adjust for coverage?

References1. A. Frazee, S. Sabunciyan, K. D. Hansen, R. A. Irizarry,

and J. T. Leek (2013). Differential expression analysisof rna-seq data at single base resolution, Biostatistics,in review.

2. L. Collado-Torres, A. E. Jaffe, J. T. Leek (2013).Manuscript in preparation.

3. https://github.com/lcolladotor/derfinder

LCT is supported by CONACyT México and LIBD.