Most accurate measures of gene expression
Date: 1.10.2014
Compared to previous methods, RNA-sequencing enables the study of alternative gene isoforms or transcripts, which are formed for example through the process of alternative splicing.
The analysis of the large amount of data produced by RNA-sequencing requires many advanced computational methods. Analysis of transcript level data is especially demanding and the differences between alternative methods can be large.
In the recent assessment the BitSeq method developed by the University of Helsinki and the University of Manchester researchers produced clearly the most reliable results in this task. In one subtask it could produce equally accurate results using only half the data needed by a very popular alternative method.
The BitSeq method is based on probabilistic modelling that allows comparing different possible origins for observed sequences that cannot be identified uniquely. This allows computing probability distributions over the expression levels of each transcript of every gene in a way that captures the uncertainty and possible sources of error in the measurements. Accounting for this uncertainty through probabilities is essential for the accuracy of the method, Antti Honkela describes.
Author: Antti Honkela, University of Helsinki
Share this article:
The assessment article:
SEQC/MAQC-III Consortium. (2014). A comprehensive assessment of RNA-seq accuracy, reproducibility and information content by the Sequencing Quality Control Consortium. Nature Biotechnology. doi:10.1038/nbt.2957http://www.nature.com/nbt/journal/vaop/ncurrent/full/nbt.2957.html
Description of the BitSeq method:
Glaus, P., Honkela, A., & Rattray, M. (2012). Identifying differentially expressed transcripts from RNA-seq data with biological variation. Bioinformatics: 28(13), 1721. doi:10.1093/bioinformatics/bts260http://bioinformatics.oxfordjournals.org/content/28/13/1721
Additional information:
Academy Research Fellow Antti Honkela, Helsinki Institute for Information Technology HIIT, Department of Computer Science, University of Helsinki, p. 050 311 2483, antti.honkela@hiit.fi
With regards, Minna Meriläinen-Tenhu, Press Officer, minna.merilainen@helsinki.fi
