Phaistos

Phaistos performs Markov chain Monte Carlo (MCMC) sampling to explore protein conformational space and support prediction and Bayesian inference of protein structures from sequence and experimental data using probabilistic models and physical force fields.

Key Features:

• Probabilistic Models: Implements probabilistic local-structure models and efficient Monte Carlo move algorithms as alternatives to conventional fragment and rotamer libraries for conformational sampling.
• Force Fields: Supports the PROFASI force field and OPLS-AA/L, with the latter including the generalized Born surface area solvent model for solvation effects.
• Bayesian Inference: Enables Bayesian inference of protein structure by integrating sequence data and experimental inputs into the sampling framework.
• Extensible Architecture: Provides an extensible computational architecture to incorporate additional move sets, probabilistic models, or force-field components.

Scientific Applications:

• Protein structure simulation: Generating conformational ensembles of proteins via MCMC sampling and energy evaluation.
• Protein structure prediction and inference: Predicting and inferring protein structures from sequence and experimental data using probabilistic and Bayesian approaches.
• Protein dynamics and interactions: Exploring protein dynamics and interaction-relevant conformational states through sampled ensembles and force-field assessments.

Methodology:

Uses Markov chain Monte Carlo (MCMC) sampling with efficient move algorithms and probabilistic local-structure models; evaluates energies with PROFASI or OPLS-AA/L (including the generalized Born surface area solvent model) and supports Bayesian integration of sequence and experimental data.