Prediction of protein–protein interactions by combining structure and sequence conservation in protein interfaces

Authors: A. Selim Aytuna, Attila Gursoy∗ and Ozlem Keskin∗ 2005
Basic idea: If A interacts with B and a's interface resembles A's and b's resembles B's, then predict a interacts with b.
Input: "template dataset" with known interacting interfaces; "target dataset" with interfaces we want to predict.

"The template dataset handles structure
and sequence conservation by combining two previously generated datasets:
the structurally non-redundant dataset of protein–protein interfaces extracted
from the PDB and the set of conserved residues on these interfaces (computational
hotspots). The target dataset is a sequentially non-redundant set of
all protein complexes and chains in the PDB."

"Proteins associate through binding sites. These sites are believed to
contribute to the biomolecular recognition and binding of proteins
by providing specific chemical and physical properties necessary
for these processes. "