Alignment PE reads with different length by TopHat

[Trudy]

Hello!
I want to know if TopHat 1.2.0 can work with paired-end reads of different lenght without problems.
Can I modify the -r/--mate-inner-dist for correct output?

Example: I have sample with insert size of 200bp, the first read of 75bp and the second read of 50bp (for quality decrease).
Can I modify -r option as 200-(75-50)? is correct?

Are there other TopHat options that I must to consider?

Thanks in advance

Valeria

[krobison]

The TopHat manual explicitly warns against this. The recommended procedure is to merge BAM files downstream with the different sizes.

Are you really truncating reads based on quality? If so, you will be stuck breaking them into separate files by length.

An option I would consider, but have not tried, is to set the quality values to 0 instead of actually trimming the data. I'm not sure whether TopHat would really effectively ignore the data if that is done.

[Gus]

So I am also interested in this. To be exact, the manual specifically warns against using different "types" of reads:

NOTE: TopHat can align reads that are up to 1024 bp, and it handles paired end reads, but we do not recommend mixing several "types" of reads in the same TopHat run. For example, mixing 100bp single end reads and 2x27bp paired ends into the same TopHat run will give bad results.

Their example only illustrates that its bad to mix paired and un-paired. It doesn't mention using a PE lib that has had non-uniform quality trimming. Does anyone have knowledge/experience with this specific case?

Gus

[selen]

Anybody tried to use different length reads for R1 and R2 for aligning their PE data? Any concerns tophat may not be able to handle?

Due to quality issues I had to trim the last 50bp of a R2 of a 100bp PE run, but R1 are still 100bp.

Thanks

[dpryan]

Anybody tried to use different length reads for R1 and R2 for aligning their PE data? Any concerns tophat may not be able to handle?

That'll work fine. After trimming, this scenario isn't infrequent.