Tophat --max-intron-length

[abisko00]

Hi!

I would like to revive this thread, as I am very much interested in an answer. Most alignment programs require this parameter and they all use different default settings. I would assume that for each genome a reasonable setting should exist. I appreciate the advice to determine it myself, but I am for sure not the only one aligning reads to e.g. the human genome. So is there any source of reasonable values for frequently used genomes?

Cheers,
Markus

[cascoamarillo]

Thats a really good idea. I haven't too much experience with tophat and RNA-seq and also finding these massive introns I wondered why? Maybe the default parameter in --max-intron-length (500000) is a little bit high, and not quite realistic (?).

[john_nl]

Maybe it would be a good idea to analyze the distribution of intron lengths in the genome you study. Then pick the 95th percentile or something like that as the maximum intron size to avoid 'odd' behavior.

[chadn737]

Does anyone have any recommendations on setting the --max-intron-length on tophat?

Few people seem to adjust this or at least report it, given the literature, and in the past I have never messed with it. However I have noticed that I often get reads, spanning dozens of genes, with massive "introns" of hundreds of thousands. My first thought is that these may be chimeric reads, but a lot of them have secondary alignments that seem correct, albeit with lower mapping quality due to mismatches.