Invariant features for searching in protein fold databases

Abstract

The tremendous growth of 3D data models available on the Internet requires more skills for fast retrieval and classification algorithms. In particular, the problem of finding structural similarities between proteins automatically, in order to predict their functional similarity, is a challenging task. In this paper a new algebraic method for structural comparison between proteins based on invariant features computed by group integration with spherical harmonics and D-Wigner matrices is proposed. Our goal is to achieve good classification without alignment by using intrinsic, pose invariant features. We compare our method to DALI, PRIDE and the Gauss Integral method in a classification and search task. Additionally we provide a Web interface to test the proposed method.

Keywords:

• Group integration
• Invariants
• Protein search and retrieval
• Protein structure
• Spherical harmonics

Acknowledgements

The authors would like to thank the anonymous reviewers for their valuable comments.

Notes

^†We write δ _y (x) for the usual δ(x−y) of the delta distribution due to space considerations.

^† http://www.ebi.ac.uk/dali/

^† http://www2.mat.dtu.dk/people/Peter.Roegen/Gauss_Integrals.html