7. Output files - devillemereuil/bayescenv GitHub Wiki

The output file is organised as in the following example:

    PEP_g   qval_g   g       PEP_alpha qval_alpha  alpha  fst
1  0.9986 0.994307 0.00041904 0.98120 0.90771 0.0060684 0.12077 
2  0.99640 0.98979 0.0019160 0.98460 0.93036 0.0032885 0.12062 
3  0.99940 0.99480 0.00013500 0.93059 0.76031 -0.061222 0.11575 
4  0.99440 0.98430 0.0023445 0.98560 0.93720 -0.0035786 0.11994 
5  0.95659 0.93859 0.038128 0.95759 0.83297 0.027243 0.12853 

The 3 first columns are related to the tests of local adaptation (g parameter), whereas the 3 following columns are related to the locus-specific effects (alpha parameter). The last column yields the estimated Fst for each locus averaged over all populations.

BayeScEnv yields two kind of test statistics:

• The PEP (Posterior Error Probability) is the estimated posterior probability the model including the parameter (g or alpha) is "wrong". The lesser it is, the more confident we can be in having found a positive result.
• The q-value is a less conservative test statistics which is averaging over the PEPs of all loci that have a lesser PEP than the focal locus. As a consequences, the q-values are generally lower than the PEP for all loci (except for the smallest one). For more information, see the publication.

The software also outputs information regarding the effect size of each parameter (g and alpha).