The dN/dS ratio shows a significant difference between the foreground and background sets. Why is the result still that selection is not associated with the phenotype/trait?

uyghbo commented 3 weeks ago

Hi team,

When using BUSTED-PH/contrast-fel to detect directional selection pressure, I found the overall dn/ds values are significantly differed in foreground and background genes. However, the results show no evidence of directional selection pressure.

The gene I tested is a drug-resistance-related gene. The foreground strains are exposed to antibiotics.

examples:

Obtaining the global omega estimate based on relative GTR branch lengths and nucleotide substitution biases

Log(L) = -451.54, AIC-c = 1127.25 (111 estimated parameters)
1 partition. Total tree length by partition (subs/site) 0.006
non-synonymous/synonymous rate ratio for background = 17.0016
non-synonymous/synonymous rate ratio for test = 0.8900

Improving branch lengths, nucleotide substitution biases, and global dN/dS ratios under a full codon model

Log(L) = -451.49, AIC-c = 1127.16 (111 estimated parameters)
non-synonymous/synonymous rate ratio for background = 23.2139
non-synonymous/synonymous rate ratio for test = 0.8081

Branch-site unrestricted statistical test of episodic diversification and association with phenotype/trait [BUSTED-PH]

Likelihood ratio test for episodic diversifying positive selection on test branches , p = 0.5000. Likelihood ratio test for episodic diversifying positive selection on background branches , p = 0.2752. Likelihood ratio test for differences in distributions between test and background , p = 1.0000.

Analysis summary (p = 0.05)

There is no evidence of episodic diversifying selection on test branches; selection is not associated with phenotype/trait

uyghbo commented 3 weeks ago

Dear Dr. Sergei L Kosakovsky Pond and Team,

When I use contrast-fel to detect directional selection pressure at the site level, I do find many sites with differences between the foreground and background. However, the BUSTED-PH method detects very few differences at the gene level. Can I use the genes with site-level differences detected by the contrast-fel method to perform functional annotation (such as KEGG/GO) to characterize the pathway differences targeted by the selection pressure in the foreground and background?

spond commented 2 weeks ago

Dear @uyghbo,

Based on what I see, you have a very small, low divergence (<0.01 subs/site) alignment. You probably also have very few branches in the foreground/background test. This means that point estimates of ω are very imprecise and are not significantly different from each other.

Can you elaborate what you see with contrast-fel? My guess would be that there should be no significantly different sites there either.

Best, Sergei

veg / hyphy