Protein-protein interaction site prediction based on conditional random fields

Minghui Li*, Lei Lin, Xiaolong Wang, and Tao Liu

Bioinformatics Research Group, ITNLP Lab,
Department of Computer Science and Technology, Harbin Institute of Technology,
No. 92 West Da-Zhi Street, Harbin, 150001, China

Dataset:

1. non-redundant hetero-complex chains
1276 chains (312858 residues) were selected from hetero complexes in the PDB (July, 2005). download

2. 1276 chains are divided randomly into three parts:
part1, part1 and part2

3. protein-protein interaction sites on hetero-complex chains
(a) sites calculated from the largest patch (download)
(b) sites calculated from the all partners (download)
(C) sites within minor interfaces (download)

4. File format

Color figures:
Fig. 4. 1W1W:E (a) CRFs, (c) SVMs and (b) Actual interface residues.
Fig. 5. 1FJG:F (a) CRFs, (c) SVMs and (b) Actual interface residues.
Fig. 6. 1KTK:A (a) CRFs, (c) SVMs and (b) Actual interface residues.

Back to homepage of Li Minghui