Pepstr is a de novo method for prediction of tertiary structure of small peptides. It uses 3 steps:
- In first step, the regular secondary structure states (helix, β-strand and coil) and β-turn types are predicted using BetaTurns method.
- Secondly, initial or starting conformation for a given sequence is generated using the Tleap module of Amber version6 with phi, psi values corresponding to the secondary structure states predicted in step I. The side chain angles are asigned using standard Dunbrack backbone dependent rotamer library.
- This initial conformation is then subjected to energy minimization and molecular dynamics simulations using Sander module of Amber. It consists of few initial cycles of steepest descent minimization followed by dynamics which is carried out for 25ps at 300K using 1-fs time steps. Finally, this is followed by minimization using a combination of steepest descent and cojugate gradient algorithms.
- The final coordinates are saved in PDB format.
The following figure is the flowchart of the prediction method used in Pepstr: