Dr. Edina Rosta
King's College London
Important biological processes taking place on the millisecond to second time scales are too slow to model using unbiased atomistic simulations. To obtain free energy profiles on long time scales, enhanced sampling methods have been developed. We show that novel Markov modelling-based tools can be used to analyse biased and unbiased simulations. Using our dynamic weighted histogram analysis method (DHAM), systematic errors due to insufficient global convergence can be corrected . In addition, DHAM also provides direct kinetic information on the conformational transitions intrinsic to the system. We also demonstrate that a variationally optimal kinetic coarse graining allows us to obtain Markov models, where not only metastable, but also transition states can be automatically identified . Applications include analysis of molecular dynamics simulations of RAF kinases  and umbrella sampling quantum classical QM/MM simulations of catalytic reactions .