BranchSite — Bayesian branch-site test for positive
selection
Usage
BranchSite(submodel: Double
-> CTMC<Codons<a>>, fs: List<Double>, omegas: List<Double>, posP: Double, posW: Double, posSelection: Int, branchCats: IntMap<Int>) → DiscreteDist<CTMC<Codons<a>>>
Arguments
A default beginning with ~ specifies a prior
distribution.
-
submodel: -
Codon model as a function of omega
-
Default:
|w:GTR +> MNM +> MutSelAA +> dNdS(omega=w)| -
fs: -
Background category proportions; one entry for each omega plus the final neutral category
-
Default:
~SymmetricDirichlet(2,1) -
omegas: -
Background omega values, excluding the final value fixed at 1; length(fs)-1 entries
-
Default:
~IID(1,Uniform(0,1)) -
posP: -
Fraction of sites switching to the foreground omega value
-
Default:
~Beta(1,10) -
posW: -
Foreground omega for switching sites when posSelection=1; ignored when posSelection=0
-
Default:
~LogGamma(4,0.25) -
posSelection: -
Model indicator: 0 for a switch to neutrality, 1 for a switch to omega=posW
-
Default:
~Bernoulli(0.5) -
branchCats: -
Branch categories: 0 for background, 1 for foreground. By default, read from tree annotations such as [&foreground=1]; unlabelled branches use 0
-
Default: The branch categories in the current context
Original default expressions
-
branchCats -
get_state(branch_categories)
Description
A Bayesian comparison based on branch-site model A. Background branches have, by default, a conserved category with 0 < omega < 1 and a neutral category with omega=1. A fraction posP of sites switches on foreground branches to omega=1 under the null (posSelection=0), or to omega=posW under the alternative (posSelection=1). The default prior places posW above 1 and gives each hypothesis probability 0.5.
The default background uses GTR, MNM, and MutSelAA. Here omega scales nonsynonymous rates after accounting for amino-acid fitnesses. Thus omega=1 retains these fitness effects; neutrality and positive selection are assessed relative to this background.
The posterior probability that posSelection=1 is the posterior probability that some sites undergo positive selection on the foreground branches.
The posterior mean of PrPosSelection can provide a more accurate estimate of this probability than the posterior mean of posSelection. The statreport tool uses LogOddsPosSelection to report the corresponding posterior log odds accurately, even when the probability is extremely close to 0 or 1. (Do not average LogOddsPosSelection directly.)
Citation
Zhang, Jianzhi; Nielsen, Rasmus; Yang, Ziheng (2005). Evaluation of an improved branch-site likelihood method for detecting positive selection at the molecular level. Molecular biology and evolution 22(12): 2472--2479. DOI: 10.1093/molbev/msi237