Analysis of protein dynamics with Markov models.
Max Linke · MPG.PuRe (Max Planck Society) · 2014
Markov State Models (MSM) are a method to find slow dynamics in proteins by approximating the slow dynamics with a Markov chain on a discrete partition of a sub-space of the configuration space.MSMs can extract this slow dynamics information from an ensemble of short simulations.To date MSMs require prior knowledge about the specific protein examined to select a sub-space and a total simulation time in the millisecond range.There have been first steps to use time-lagged independent component analysis (TICA) [1] to automatically find the slow sub-space in a protein.TICA has been recently used [2] with a 30 residue intrinsically disordered peptide kinase inducible domain.We found that TICA is not guaranteed to always find the slow sub-space in a MD-simulation.We could also show that TICA can be used to extract slow dynamics information with MSMs from 100 Ubiquitin simulations with a total simulation time of just 38 µs.iii