Protein Co-evolution and Its Applications in Residue Contact Prediction
Fangfang Pan1 *, Dongsheng Che2 *, Michelle Momany 1 , Liming Cai2, Russell L. Malmberg1
1Department of Plant Biology, University of Georgia, Athens, GA 30602, USA
2Department of Computer Science, University of Georgia, Athens, GA 30602, USA
*These authors contributed equally to this work.
Predicting protein 3D structure from primary sequence is a fundamental problem in computational biology, and building a residue physical contact map is a key step. The physical interactions between residues determine protein structure and functions, and those often involve highly conserved residues detected from homologous sequences. However, residues can also mutate in a correlated way to preserve important interactions. Residues’ co-evolution feature may imply that these are important residues under interaction constraints to keep function and structural stability. Thus, co-evolving residues are one special group of residues worth studying for physical interactions. The increase in the number of available protein sequences and structures allows the application of the mutual information statistic to protein co-evolution.
We focused on 48 pairs of interacting proteins from the Protein Complex Crystallization Database and used mutual information to study residue co-evolution. Our analysis showed that, on average, co-evolving residue pairs were physically closer to each other than those in the background. We derived protein co-evolving residue preference matrices using those co-evolving residues, and observed the existence of some individual residue pairing preferences. We also derived contact scoring matrices for co-evolving residues, and developed a new method to computationally predict co-evolving residue pairs to be proximal or distant. Preliminary prediction results indicate the potential application of our approach in protein structure analysis.
- Source Code: coevolve-miv0.1.tar.gz (Linux version)
Dongsheng Che and Russell Malmberg
Last updated 06/20/2007