Transcription Start Sites (TSS) analysis for 5' RNAseq data

[liux]

I recently start working on a TSS analysis project (yeast). And want to get some suggestions.

The experiment had mRNA from two population of yeast sequenced on 5' only. The goal is to detect the TSS and compare the usage between them. From the alignment it is clear that reads are piling up upstream of genes, sometimes in more than one piles of the same gene.

I first went off to find existing software for analyzing this data. Homer has a nice page about how to use ChIPseq methods for TSS data. I used it to call 'peaks'. when overlapping with the alignment data, it is clear that Homer missed a lot of secondary TSS, which could be problematic.

While Homer treats the TSS 'peak' as IP peak, they are quite different. For one, TSS 'peak' are very narrow, if you only consider the beginning of the reads is the TSS. unlike ChIP peaks are composed of reads representing fragments that have binding site somewhere on the fragment, the 5' RNAseq reads have the TSS right at the begining of the reads.

I am wondering if there are TSS analysis tool for 5' RNAseq.

Also if I want to implement a method similar to ChIPseq, but instead of extending read to ~200bp fragment, but only use the 5' tip of the read, which method would be simple enough for quick proof-of-concept run?

Thanks for all your suggestions!

XyL

[syfo]

That sounds like CAGE. I suggest you take a look at the work from Piero Carninci's group (RIKEN), in one of their papers I think they identified several types of TSS "peaks". Look for Timo Lassman, for instance: MOIRAI: a compact workflow system for CAGE analysis

Also, it would help to give more information about the actual experiment because the relevance of the bioinfo analysis depends on the upstream protocol (ChIP-seq and RNA-seq are quite different experiments).

As for extracting the 5' position of mapped reads I'd use BEDtools.