We show that a single-copy tempering method is useful in protein-folding simulations of large scale and high accuracy (explicit solvent, atomic representation, and physics-based potential). The method uses a runtime estimate of the average potential energy from an integral identity to guide a random walk in the continuous temperature space. It was used for folding three mini-proteins, trpzip2 (PDB ID: 1LE1), trp-cage (1L2Y), and villin headpiece (1VII) within atomic accuracy.
Further, using a modification of the method with a dihedral bias potential added on the roof temperature, we were able to fold four larger helical proteins: α3D (2A3D), α3W (1LQ7), Fap1-NRα (2KUB) and S-836 (2JUA).
We also discuss how to optimally use simulation data through an integral identity. With the help of a general mean force formula, the identity makes better use of data collected in a molecular dynamics simulation and is more accurate and precise than the common histogram approach.
| Identifer | oai:union.ndltd.org:RICE/oai:scholarship.rice.edu:1911/71707 |
| Date | 24 July 2013 |
| Creators | Zhang, Cheng |
| Contributors | Ma, Jianpeng, McNew, James, Igoshin, Oleg |
| Source Sets | Rice University |
| Language | English |
| Detected Language | English |
| Type | thesis, text |
| Format | application/pdf |
Page generated in 0.0018 seconds