The Block Gibbs Sampler for RNA Prediction (RNAG)
RNAG is an application for the prediction of the RNA secondary structure common to multiple unaligned sequences (Wei et.al. Bioinformatics,27:2486-2493 2011). It uses a blocked Gibbs Sampler strategy that has has a theoretical advantage in convergence time (Liu, J Amer. Stat. Assoc,89:958, 1994). This Bayesian sampling algorithm returns structural predictions using gamma centroid estimators (Hamada, et.al. Bioinformatics, 25:465, 2009), which allow the user to adjust the sensitivity specificity trade off. Sampled structures provide for a more complete characterization of posterior space in two ways: a clustering of sampled structures for the evidence of mufti-modality of the posterior space, and assessment of the uncertainty of predictions using Bayesian confidence limits (Webb-Robertson, PloS Comp Bio, doi:10.1371/journal.pcbi.1000077, 2008).