Hi everyone,
Has anyone successfully calculated delta K using the methods in Evanno et al using fastSTRUCTURE output, as opposed to STRUCTURE output, and if so, what did you use as the equivalent of the required raw data value of "LnP(D)" that is supposed to be generated in the output?
A little more detail: I'm trying to calculate delta K as outlined in Evanno et al (https://www.ncbi.nlm.nih.gov/pubmed/15969739). This requires taking the posterior probability of the data for a given K, Pr(X|K) as a raw data value. The Evanno paper tells me that STRUCTURE gives this value in the output and calls it ‘Ln P(D)’.
I cannot find an equivalent to Ln P(D) in fastSTRUCTURE output, though. There is no kind of probability value for each K value in the output. I've pasted a sample output log below. My K range is 1 to 10. Here is what the output looks like:
Marginal likelihood with initialization (1) = -0.2905800557 Marginal likelihood with initialization (2) = -0.2907194453 Marginal likelihood with initialization (3) = -0.2934433316 Marginal likelihood with initialization (4) = -0.2868424888 Marginal likelihood with initialization (5) = -0.2846523158 Iteration Marginal_Likelihood delta_Marginal_Likelihood Iteration_Time (secs) 0 -0.2846523158 -- 818.690 10 -0.2843030467 0.0003492690 113.575 20 -0.2842688896 0.0000341571 125.511 30 -0.2842612894 0.0000076002 120.150 40 -0.2842591377 0.0000021517 115.506 50 -0.2842581353 0.0000010024 112.907 60 -0.2842574538 0.0000006815 117.128 Marginal Likelihood = -0.2842574538 Total time = 1524.7450 seconds Total iterations = 60
