The 'baldur' package.

Description

baldur is a Bayesian hierarchical model for statistical decision in proteomics data. It models the mean-variance trend with the option of two different regression models, a gamma regression or a latent gamma mixture regression. It then the regression model as en Empirical Bayes estimator for the prior on the variance. Further, it assumes that each measurement has an uncertainty (increased variance) associated with it that it also infers. Finally, it tries to estimate the posterior distribution (by Hamiltonian Monte Carlo) for the differences in means for each peptide in the data. Once the posterior is inferred, it integrates the tails to estimate the probability of error from which a statistical decision can be made.

References

Berg and Popescu (2023). Baldur: Bayesian Hierarchical Modeling For Label-Free Proteomics Exploiting Gamma Dependent Mean-Variance Trends. Under review. Stan Development Team (2022). RStan: the R interface to Stan. R package version 2.21.5. https://mc-stan.org