Hi all,
I have raw count RNA seq NGS data, I want normalize it and apply machine learning and data mining algorithms on it. I want to know that, what are the bests or suitable normalization methods are specific for this kind of data?
•
written
by
jack •
A recent paper has evaluated different normalization methods. A comprehensive evaluation of normalization methods for Illumina high-throughput RNA sequencing data analysis http://bib.oxfordjournals.org/content/14/6/671.full.html?etoc
The conclusion of this paper is basically DESeq or EdgeR methods (TMM) were the best, and fairly similar.
Recently we had some data that was normalized much better with TMM/EdgeR than DESeq, so you may want to try both.
imho, they are both quite similar since both of them use negative binomial distrib. for modelling. If you are not satisfied you can also check out DEGSeq. It uses poisson, we used it for first analysis and found that it is kind of more conservative. Switched to DESeq, though.
This recent comparison of methods paper for RNA-seq also covers normalization: http://www.ncbi.nlm.nih.gov/pubmed/23497356
Log in to answer this question.
Check Optimal Scaling of Digital Transcriptomes
Hi jack, I am doing similar works as you do before,and I also meet the problem of which methods to adopt for normalization of different kinds of RNA-seq count for a subsequent machine learning .And have you get the solution?Could you give me some advices?