RNAseq transcriptome analysis: workflow problem

[lollysticky]

Hi all,


I will be performing RNAseq transcriptome analysis on a certain organism under a specified number of conditions (let's say 10). The goal is to construct a tab-delimited file which contains the expression values (the raw read counts, not RPKM/FPKM values) for each gene under all conditions.
I am the first one in the lab to perform such an experiment and we don't have a standard workflow developed. Therefore I would ask you, the community, to review what I sketched so far and respond to my questions if possible!

my design so far:
1) generate RNAseq data
2) preprocessing the data: FASTX-toolkit (quality check, trimming, clipping, filtering)
3) aligning the reads -> TopHat (SAMformat output)
4) Iterate 1-3 for each condition...
5) construct file
6) further analysis


Questions:
a) what to do with isoforms? Do I take them into consideration (using Cufflinks or so), or not? My organism has very few introns, and I expect to see little isoform transcripts. Nevertheless, any isoform information is valuable.
b) how to tackle multireads? As far as I understand it, TopHat does not carry out some multi-read re-distribution like ERANGE does...
c) how to proceed to raw read counts? TopHat reports RPKM values, but I need raw read counts -> I could use some sort of comparison script which uses my annotation files to construct a read count for each gene? (BEDtools can do this I think).

If anybody has a better suggestion for a workflow and/or possible answers to my questions, please post them here


thanks!

[john_mu]

SpliceMap does separate multi-reads, you could check the paper to see if it suits your purpose. Although the paper does describe an older version, it gives the main idea.

The link is in my signature.

[lollysticky]

We have used TopHat in the lab before and would like to continue using it
that's why I'm searching for a solution

[frozenlyse]

One thing to note - TopHat will only work with sequences of all the same length, so trimming adaptors is a bit of a *****

[jameslz]

One thing to note - TopHat will only work with sequences of all the same length, so trimming adaptors is a bit of a *****

I met the same situation, but it sees to be better if trimming the low quality base(such as 'B') from the 3' end.