These tests were calculated with 1,000 bootstrap replicates.
ML branch lengths of alternative topologies were first inferred assuming a concatenated WAG+F+Γ4 model using Tree-Puzzle [33], site-wise log-likelihood values were then computed with Codeml [37] and p-values of the different likelihood-based tests were calculated with Consel.
First, culture was used as the gold standard reference, and prevalence, sensitivities, specificities, positive and negative predictive values (PPV's and NPV's) for the four serological tests were calculated with exact 95% confidence intervals using the Stata 10.1 statistical software package (Stata Corp., College Station, Tex).
Fisher tests were calculated with the fisher.test function of the R statistical software [ 32].
Sensitivity and specificity for both tests were calculated with respect to clinical outcome.
All tests were calculated with the SigmaStat software (Jandel Scientific, San Rafael, CA, USA) JV, GN and NR conceived the experimental design.
In the present analyses, these tests were calculated with the alternative software Mplus using the weighted least squares means and variance adjusted estimation method (WLSMV) [ 64].
These subgroup differences (as well as ANOVA post-hoc tests) were calculated with non-parametric Mann–Whitney U-tests reporting exact significance values, since assumptions for t-tests were not fulfilled.
The maximum pressure in the closed bomb test was calculated with modified SE-CSP virial coefficient expressions with the calculated error of less than 2%, and the error was smaller than the result calculated with the reported values under the same calculation conditions.
However, the overall damage in the line test was calculated with a laboratory bench calibration coefficient.
