A recursive connectionist approach for predicting disulfide connectivity in proteins

ABSTRACT

We are interested in the prediction of disulfide bridges in proteins, a structural feature that conveys important information about the protein conformation and that can therefore help towards the solution of the folding problem. We assume here that the disulfide bonding state of cysteines is known and we focus on the subsequent problem of disulfide bridges pairings assignment. In this paper, disulfide connectivity is modeled by undirected graphs. A graphspace search algorithm is employed to explore alternative disulfide bridges patterns and prediction consists of selecting the 'best' graph in the search space. The core of the proposed method is a recursive neural network architecture trained to score candidate graphs. We report experiments on previously published data showing that our algorithm outperforms the known alternative methods for most proteins. Furthermore, we assess the generalization capabilities testing the model on previously unpublished data.

REFERENCES

• 1: A. Bairoch and R. Apweiler. The SWISS-PROT protein sequence database and its supplement TrEMBL. Nucleic Acids Res., 28:45--48, 2000.
• 2: P. Fariselli et al. Role of evolutionary information in predicting the disulfide-bonding state of cysteine in proteins. Proteins, 36:340--346, 1999.
• 3: P. Fariselli et al. Prediction of disulfide connectivity in proteins. Bionformatics, 17:957--964, 2001.
• 4: P. Fariselli et al. A neural network-based method for predicting the disulfide connectivity in proteins. In Proc. 6th Int. Conf. Knowledge Engineering Sys., 2002.
• 5: A. Fiser and I. Simon. Predicting the oxidation state of cysteines by multiple sequence alignment. Bionformatics, 3:251--256, 2000.
• 6: P. Frasconi, M. Gori, and A. Sperduti. A general framework for adaptive processing of data structures. IEEE Trans. on Neural Networks, 9:768--786, 1998.
• 7: P. Frasconi, A. Passerini, and A. Vullo. A two state SVM architecture for predicting the disulfide bonding state of cysteines. In Proc. 12th IEEE Workshop on Neural Networks for Signal Processing, 2002.
• 8: B. Schoelkopf and A. Smola. Learning with Kernels. MIT Press, 2002.
• 9: W. Wedemeyer et al. Disulfide bonds and protein-folding. Biochemistry, 39:4207--4216, 2000.

INDEX TERMS

Keywords:
disulfide connectivity prediction, protein folding, recursive neural networks