Hi hs,

For two trees (as is the case in SBP), KH and SH testing approaches are equivalent. Use the 'Run 1' option. Set the number of replicates to the reciprocate of the p-value you are interested in. For example -- if you seek to establish multiple trees at p = 0.01, use 100 replicates (technically, you will get p <= 1/101 if all replicates are significant).

Sergei

In SBP, KH Test output file (with extension of 'kh') consists of 5 columns (Breakpoint, Part 1 KH p-value, Part 1 Delta LogL, Part 2 KH p-value, Part 2 Delta LogL). Is the site declared to be a recombination breakpoint if both (or either?) of 'Part 1 KH p-value' and 'Part 2 KH p-value' are <0.05?

For some gene, 'There seems to be NO recombination in this alignment' based on AIC, AIC-c, and BIC, while KH Test (BOTH p-values are <0.05) identified several breakpoints. For other gene, a position of best supported breakpoint based on AIC, AIC-c, and BIC was different from any breakpoint positions identified by KH Test.

2). Do not use small p-values for KH tests to assess significance.
Does 'small p-values' mean small significance levels (e.g. 1% and 0.1%)?

SBP printed the following error when selected '(2):[Run 1]' for 'KH Testing', while SBP worked well when selected '(1):[Skip]' for 'KH Testing'.
##########
Error:
Cannot reuse the filter 'filteredData1' because it is already being used by likelihood function 'lf2', and the two likelihood functions impose different
leaf-to-sequence mapping. Create a copy the filter and pass it to the second likelihood function to resolve this issue.

Function call stack
1 : Construct the following likelihood function:filteredData0 ; givenTree0 ; filteredData1 ; givenTree1 ;
-------
2 : RunASample(individual)
-------
##########