Hi All,

I am hoping for some clarification on a PARRIS output. The test is of a gene that we think is under pretty strong purifying selection.. Several negatively selected codons are identified, and only 1 codon indicated as being under diversifying selection. This result is what we had expected, knowing something about the function of the gene..

However, PARRIS (see attached screenshot) tells us that there is significant positive selection, and Im assuming the positive=diversifying.  Interestingly, the mean  dN/dS=0.7, suggestive of purifying selection. So, what am I missing here?

Thanks, Matt

---

Hi Matt,

PARRIS seems to indicate that about 20% of the sites in the gene are under positive (diversifying) selection with omega = 3.2 (under inferred rate distributions for the alternative model). On average, the gene is under purifying selection (mean dN/dS = 0.7), but a proportion of sites appear to be under positive selection. If your alignment is small (e.g <30 sequences), it is common that very few single sites reach significance levels to be detected as positively selected by SLAC/FEL/REL or other site-wise approaches. PARRIS pools sites into three groups (e.g. in this case 20% with omega = 3.2 and 80% with omega = 0.09) and makes inferences about those groups. If the data set is small, there is little power to discriminate with confidence WHICH sites are under selection, but there is enough power to say that SOME sites are.

HTH,
Sergei