Hi Sergei,

I have done analysis to find evidence of positive sellection in ORF of 90-100 H5N1 isolates using datamonkey. When using IFEL and REL at least 1 positive selection found; however, FEL and PARRIS couldn't find it. Could you please give me suggestion, which one is better to find positive selection in the alignment of H5N1 isolates? And why I should choose it?

Thank you,

Regard,
Hendra  :)

Hi Hendra,

FEL and IFEL test for selection differently: FEL looks at the entire tree, while IFEL only looks at interior branches. If you have positive selection along interior branches followed by purifying selection at the tips, FEL will not pick up the same site.

PARRIS and REL also performs different tests: PARRIS is a global test (is there statistical evidence for selection  on a non-zero proportion of sites), while REL is an empirical site-by-site test. Positive REL results coupled with negative PARRIS results are indicative of weak selection signals or possible false positives returned by REL.

Best,
Sergei

Hi Sergei,

Thank you for your reply.

It does meant FEL and PARRIS is more appropriate to look for the evidence of positive selection in this case?

Many thanks,

Cheers,
Hendra

Hi Hendra,

What method is more or less appropriate really depends on your question. FEL is a good 'default' method.

Cheers,
Sergei

Hi Sergei,

Thank so much for your your help.

Cheers,
Hendra