Circular clustering of protein dihedral angles by Minimum Message Length

摘要: Early work on proteins identified the existence of helices and extended sheets in protein secondary structures, a high-level classification which remains popular today. Using Snob program for information-theoretic Minimum Message Length (MML) classification, we are able to take dihedral angles as determined by X-ray crystallography, cluster sets into groups. Previous Hunter States has applied similar Bayesian method, AutoClass, data with site position represented 3 Cartesian co-ordinates each alpha-Carbon, beta-Carbon Nitrogen, totalling 9 co-ordinates. By using von Mises circular distribution program, instead represent local properties two angles, phi psi. Since can be modelled having 2 degrees freedom, this orientation-invariant angle representation is more compact than that nine highly-correlated message length concepts discussed paper, such concise model likely underlying generating process from came. We report results our plotting classes (phi, psi) space; introducing symmetric distance measure build minimum spanning tree between classes. also give transition matrix note three region approximately -1.09 rad psi -0.75 close have high inter-transition probabilities. This gives rise tight, abundant self-perpetuating structure.