---

We present a method to predict the flexible and rigid regions based on sequence. We use the free energy of two consequent amino acids to define a factor for distinguishing flexible regions from the rigid ones. Using statistical analysis of this free energy, we assign a normalized number between zero to one hundred which we call it flexibility number. Taking the effects of up to four neighbors of an amino acid, into account, resulted in an efficient prediction of flexible and rigid regions of a protein.

---

Aims: Prediction of three-dimensional structure from a sequence of amino acids is one of the important problems in structural bioinformatics. Proteins select a special structure among many possible conformations in order of seconds. Levinthal paradox expresses that random searches could not be an effective way to form a native structure and a principal mechanism should be available. Reduced alphabet fewer than 20 have been interested in protein structure because it could sufficiently simplify the protein folding problem. It is generally assumed that the native structure form in the lowest free energy among all conformational states. Therefore, it is needed to design a trustworthy potential function that could discriminate protein fold from incorrect ones.
Materials and Methods: Knowledge-based potential functions are one type of energy functions derived from a database of known protein structures. In this study, we introduce a knowledge-based potential and assess the power of five amino acids ALA, LEU, ILE, VAL, and PHE in discrimination of native structure using the reduced model. In the reduced model only the energy between the aforementioned amino acids are calculated.
Finding: The reduced model was evaluated using four criteria. The results indicate that there is no significant difference between the 20- amino acid model and the reduced model.
Conclusion: The presented model indicates that the power of discrimination of native structure is originally from the interaction between the aforementioned amino acids. Therefore, it needed a new strategy to capture the remaining interactions to improve the power of knowledge-based potential function.