Standard errors for the observed N positive and observed N negative were estimated using a bootstrap technique to generate many sets of simulated data to which the negative binomial function was fitted and the standard error calculated for the simulated N positive and the simulated N negative for each AUDIT score.
Marker accuracy was analyzed using linear mixed effects models as implemented in the lmer-function of the lme4 package (Bates, Maechler, Bolker, & Walker, 2015) for R (R Development Core Team, 2012), using a binomial link function and the Bobyqa optimizer.
We used generalized linear models with a binomial distribution function and the canonical log link to calculate each compound probability of activity from the scores obtained by the compounds in a virtual screening experiment.
Note however that the divergence estimates made by the binomial mass function and the analysis with BEAST are conservative (i.e., the true time of the recombination event are probably more recent) given that we had only three sequences in the analysis.
For both the observed N positive and the observed N negative distributions we selected a negative binomial function to fit the data because of the discrete properties of the scale (Fig. 1a,b).
After each mitotic cycle, the n for each cell was doubled and then distributed randomly according to a binomial function between the two daughter cells.
Measures of association are presented with 95% confidence intervals calculated by use of the binomial function in the STATA statistical software version 12.
We then fitted each of the 46,130 distance distributions using a negative binomial distribution function, and generated the corresponding cumulative density functions giving the p value as a function of the city block distance (Fig. 1i).
From this framework it was possible to focus on the ligand molecules using discriminant analysis, considering the curve of the common binding affinity values and the binomial function of binding specificity per receptor.
Because these models are equally likely (Coyne and Orr, 2004), we used the random binomial function in R to assign the 207 conspecific-associated locus pairs to each fitness matrix for each simulation.
