Katarzyna Stąpor

Chaperonin structure: the large multi-subunit protein complex.

The multi sub-unit protein structure representing the chaperonins group is analyzed with respect to its hydrophobicity distribution. The proteins of this group assist protein folding supported by ATP. The specific axial symmetry GroEL structure (two rings of seven units stacked back to back - 524 aa each) and the GroES (single ring of seven units - 97 aa each) polypeptide chains are analyzed using the hydrophobicity distribution expressed as excess/deficiency all over the molecule to search for structure-to-function relationships. The empirically observed distribution of hydrophobic residues is confronted with the theoretical one representing the idealized hydrophobic core with hydrophilic residues exposure on the surface. The observed discrepancy between these two distributions seems to be aim-oriented, determining the structure-to-function relation. The hydrophobic force field structure generated by the chaperonin capsule is presented. Its possible influence on substrate folding is suggested.

Using the one-versus-rest strategy with samples balancing to improve pairwise coupling classification

Abstract The simplest classification task is to divide a set of objects into two classes, but most of the problems we find in real life applications are multi-class. There are many methods of decomposing such a task into a set of smaller classification problems involving two classes only. Among the methods, pairwise coupling proposed by Hastie and Tibshirani (1998) is one of the best known. Its principle is to separate each pair of classes ignoring the remaining ones. Then all objects are tested against these classifiers and a voting scheme is applied using pairwise class probability estimates in a joint probability estimate for all classes. A closer look at the pairwise strategy shows the problem which impacts the final result. Each binary classifier votes for each object even if it does not belong to one of the two classes which it is trained on. This problem is addressed in our strategy. We propose to use additional classifiers to select the objects which will be considered by the pairwise classifiers. A similar solution was proposed by Moreira and Mayoraz (1998), but they use classifiers which are biased according to imbalance in the number of samples representing classes.

A New Approach to Multi-class SVM-Based Classification Using Error Correcting Output Codes

Protein fold classification is the prediction of protein’s tertiary structure (fold) from amino acid sequence without relying on the sequence similarity. The problem how to predict protein fold from amino acid sequence is regarded as a great challenge in computational biology and bioinformatics. To deal with this problem the support vector machine (SVM) classifier was introduced. However the SVM is a binary classifier, but protein fold recognition is a multi-class problem. So the method of solving this issue was proposed based on error correcting output codes (ECOC). The key problem in this approach is how to construct the optimal ECOC codewords. There are three strategies presented in this paper based on recognition ratios obtained by binary classfiers on the traing data set. The SVM classifier using the ECOC codewords contructed using these strategies was used on a real world data set. The obtained results (57.1% - 62.6%) are better than the best results published in the literature.