When awaiting bio-champollion – dynamic-programming regularization of the protein secondary structure predictions

Predictions of protein secondary structure using current methods are often unrealistic, i.e. the predicted alpha-helices or beta-strands are too short. To improve the realism, various heuristic ‘filtering’ or ‘smoothing’ methods are used. They are more or less intuitive and are based on ad hoc corrections. We present a regularization method to obtain a realistic structure from predicted propensities. It is based on the known dynamic programming algorithm and is quite objective. It can be used with any prediction method which yields propensities. The regularized predictions conserve well the overall prediction accuracy and improve the ‘protein-likeness’ of the prediction.