TopHat and Multireads - what is happening under the hood?

[choy]

Hi all -

I have had success so far with TopHat, and it is a great tool!

My RNA-seq workflow is to run TopHat and then Cufflinks. Recently, I noticed that some of the transcripts that are produced are "false positives" due to multi-mapped reads. In otherwords, the vast majority of coverage (RPKM) in those transcripts due to reads that map to multiple locations.

I could eliminate all of these false positives by setting TopHat's (-g/--max-multihits) option to '1', but I would rather not do this because it would sacrifice sensitivity in exchange for guaranteed uniqueness.

Ideally, I want to be able to know some statistics about how many of the reads in each transcript came from multi-mapping vs. uniquely mapping reads. Is there any information in the 'accepted_hits.sam' file about this? What about in any of the other temporary files that TopHat generates.

How does everyone else handle multi-mapping reads and TopHat?

Also, in the SAM output file, I noticed that the mate pair reference is always set to '=' even if the two mates map to different references.

Finally, would it be possible to add the 'insert size' parameter to the accepted_hits.sam file?

Best regards, and thanks again!

[MerFer]

Hi all -
Also, in the SAM output file, I noticed that the mate pair reference is always set to '=' even if the two mates map to different references.

Finally, would it be possible to add the 'insert size' parameter to the accepted_hits.sam file?

Best regards, and thanks again!

In tophat manual there is already an option for the expected inner distance between mate pairs. I think, it would help you.

-r/--mate-inner-dist <int> This is the expected (mean) inner distance between mate pairs. For, example, for paired end runs with fragments selected at 300bp, where each end is 50bp, you should set -r to be 200. There is no default, and this parameter is required for paired end runs.

Best
Meryem

[choy]

In tophat manual there is already an option for the expected inner distance between mate pairs. I think, it would help you.

-r/--mate-inner-dist <int> This is the expected (mean) inner distance between mate pairs. For, example, for paired end runs with fragments selected at 300bp, where each end is 50bp, you should set -r to be 200. There is no default, and this parameter is required for paired end runs.

Best
Meryem

Thanks Meryem. I am actually setting this parameter already. However, even though we pass the expected inner distance to TopHat, each individual mate pair will have its own insert size when mapped.

I suppose that since TopHat maps to the human genome, it is not possible to approximate the actually insert size until transcript assembly is complete.

Best regards,