Tophat segmentation question

[rrdavis]

I am using TopHat 1.4.1 on some Paired End RNA-Seq data (2 X 40 bp).

I get the following warning from Tophat with runing the default segment-length:

Warning: you have only one segment per read.
If the read length is greater than or equal to 45bp,
we strongly recommend that you decrease --segment-length to about half the read length because TopHat will work better with multiple segments


However, in the release note for TopHat 1.3.1 it states:

"For short reads (usually <45-bp), it is recommended that users decrease segment length (--segment-length) to about half the read length and segment mismatches (--segment-mismatches) to 0 or 1"

Is it correct to assume that I should set for my example, --segment-length to 20?

It's a little confusing to have that output from Tophat. Thanks!

[DineshCyanam]

Yes, you are right. Tophat tries to align the whole reads first and then the unmapped reads are split into small segments so in your case 9 (--segment-length 20) they would be split into 20bp fragments and aligned again.