How to find alteranative TSS between two group of samples

[Fernas]

Dear all,

I have CAGE data belongs to different samples of two groups (A and B), where each group have 10 samples (biological replicates) . I want to detect the alternative TSS between the two groups. I mapped the reads using Tophat2.0.8 and wrote my own code to assign mapped reads to genes. The problem that I have is: when I view the mapped reads using IGV, I see some samples of a group have a specific TSS but other samples of the same group have another TSS. In some cases, multiple TSSs are found in different places for each sample and it is different than other samples. (Note: no problem exists in the mapped data. The data was filtered, mapped with very strict parameters...etc)

So, I am wondering what is the best strategy to find a common TSS for a group of samples? (I will use that group TSS to compare it between the two groups). I thought of two possible ways:
1) Define a common TSS to be the position which is common between maximum number of group's samples. (what if that position is inside the gene body or close to gene end position!!)
2) cluster TSS positions of each sample using some tools and then find the common cluster between group's samples.

Does anyone have better solution? any suggested plotting tool and/or visualization tool for the TSSs?

[Dario1984]

Your data is interesting. There is no CAGE-seq experiment with that many biological replicates available to the public. You can create any definition of what is the most likely TSS, as long as you can justify it.

[gringer]

I mapped the reads using Tophat2.0.8 and wrote my own code to assign mapped reads to genes. The problem that I have is....

Does anyone have better solution?

How about Cufflinks? It's the next step in the tuxedo suit and seems like a logical follow-on from Tophat that should do what you're saying you want to do, so presumably you have some other reason for writing your own code. But just in case:

http://cufflinks.cbcb.umd.edu/manual.html#cuffmerge