What Is A Good Ssr Filtering Cutoff?

Question

[MrGreen]

I have a library of repeats, found using a de novo methodology. TRF tells me that I've got a ton of SSR content within the library, with SSR composition ranging up to 90% within an element.

What would be a good percent SSR cutoff to filter out a repeat, and why?

Some figures come to mind, like 80% (Wicker et al. 2007) for repeat classification or 35% in the S. bicolor de novo repeat library (Paterson et al. 2009).

Answer 1

[SES]

Transposable elements do contain simple repeats and LTR retrotransposons in particular may contain large arrays of SSRs in the LTR regions. I wouldn't try to remove these sequences if you are constructing a repeat library because these are unique features of specific elements that will be informative. One thing to keep in mind is that tools like BLAST may be doing simple repeat filtering already (with a program called DUST) so this may not be any issue at all (depending on what you are trying to do).

The numbers you mention in the last statement refer to thresholds for defining TE families, and this is not related to "filtering out" repeats.