Running TopHat with 32bp or 36bp reads

[statsteam]

Hi all,

I am trying to run TopHat with 32bp or 36bp reads. I am setting the length of segments to the half of the original length (16bp for 32bp reads and 18bp for 36bp reads). Besides it takes too long to map such short reads to the genome, I am not so sure about the reliability of the results because there are going to be simply too many mappable hits on the genome.

Any thoughts on this?

Thank you,
Statsteam

[HTS]

Further breaking down reads as short as 32 or 36bp doesn't seem to be a sensible thing to me. I wouldn't use segments shorter than 25bp myself. Doesn't the default setting more or less work for you in your case? I used the -a 5 option with 32bp reads and things seem to work fine for me. To further increase sensitivity, you could supply known junctions via a GFF3 file or tune down the -F parameter.

BTW, I saw in your other post that you have worked with Burge's cerebellum data. Where did you get it? By following the accession numbers provided in their Nature paper, I get all but the cerebellum data of six individuals. Thanks!

-- Leo