Discordant results between FEL and MEME on Datamonkey

[katherinermartin]

I'm finding that when I run FEL and SLAC on HyPhy on Datamonkey, I get conflicting results for the same dataset. For site 16, FEL identifies it as evolving under purifying selection, but MEME identifies it as a positively selected site. For both FEL and MEME, I used p-value threshold = 0.05. For FEL, I provided the user defined tree. Attached are the .json for each analysis, happy to provide alignments/trees. One thing to note is that this is a rather short alignment, as the whole alignment had a breakpoint, so this actually represents a partition after that breakpoint. part2poly_MEME.json part2poly_FEL.json

[spond]

Dear @katherinermartin,

These types of disagreements are expected (see https://journals.plos.org/plosgenetics/article/info%3Adoi%2F10.1371%2Fjournal.pgen.1002764)

FEL looks for pervasive (i.e. on average) selection, while MEME looks for episodic selection. So your site 16 is on average under negative selection, but there are episodes of positive selection as well. See, in the figure below, that site 16 is quite variable, with many synonymous AND non-synonymous subs.

Best, Sergei

[katherinermartin]

Ah okay that makes a ton of sense. Thank you so much @spond 😄