Çiğdem Sevim
PhD, Koç University
Finding Lowest Energy
Conformations of Denatured Proteins
Using Hidden Markov
Models
In targeted drug design, knowledge of the
highly probable conformations of a given peptide
that is designed to recognize, bind to, and block
the activities of proteins is important. The
conformation of the peptide has to be the most
probable conformation for maximum stability.
Based on the Hidden Markov Model (HMM)
framework, the problem of finding the state
sequence in the HMM that maximizes the
probability of a conformation given the observed
sequence can be calculated. A dynamic programming
algorithm, called the Viterbi algorithm
can be used to solve the problem. The pairwise
dependent probabilities for a given sequence
are obtained from protein libraries. Introducing
these priori probabilities, we may obtain
pairwise dependent torsion angles using rotational
isomeric states (RIS) model in which
each molecule, or bond is treated as occurring in
one or another of several discrete
rotational states.