Tag-Seq Data Analysis

[Tagz]

Hello!

Can anybody help me with informations concerning the analysis of Tag-Seq (Tag-Profiling/Illumina) data? I´m new to this topic and already posted a thread in the bioinformatic section - no reply so far.

Are there already packages available and if this is the case which one is best to use? Or is it possible to use standard programs for RNA-Seq (like ERANGE)?
What kind of steps need to be implied?
I´ve read about an in-house perl script (Morrissy et al. 2009). Does anybody have an equivalent? Is it possible to test it?

I would appreciate any advice.
Thanks a lot.

[ybz]

Hi Tagz,

It's possible to use the DiscoverySpace platform to analyze Tag-seq data (http://www.bcgsc.ca/platform/bioinfo/software/ds). It was originally developed for LongSAGE data, so there is one caveat when working with the much much larger Tag-seq libraries: you have to remove the low frequency tags (<5 counts) to make the libraries small enough to be manageable by the software. Statistically speaking, tags occurring at a frequency of <5 are not statistically very different than 0, so it is not a huge compromise, depending on what your analysis is. For instance, if you're looking for statistically significant changes in tag expression between two libraries, having a count of 200 in one library and either 5 or 0 in another library would give upi pretty much the same result.

Alternatively, if you have time to wait, some of the in-house scripts mentioned in the paper will be made available as part of a second publication coming out in the next couple of months.

There may be other tools out there for analyzing LongSAGE data that could be used for Tag-seq data, but I have not looked for them.

best of luck

[kmcarr]

See this thread:

http://seqanswers.com/forums/showthread.php?t=498

[Davis McC]

DE analysis of Tag-seq using edgeR

Hi Tagz

The Bioconductor package 'edgeR' can be used to analyse Tag-seq data for differential expression (http://bioinformatics.oxfordjournals...hort/26/1/139; http://www.bioconductor.org/packages...tml/edgeR.html). Indeed, we have achieved good results when using the package to (re-)analyse publicly available Tag-seq data (e.g. from T Hoen et al 2008, doi:10.1093/nar/gkn705).

Cheers
Davis

[Simon Anders]

Hi

in my experience, the first step will be to visualize your alignment in a genome browser to see how many tags per transcript you get. (Don't count on there being only one.) Then, you can create a count table, e.g. with our htseq-count script, and analyse this count table with either edgeR (Davis et al.'s package, see post #4) or DESeq (our package).

Simon