On modeling protein superfamilies with low primary sequence conservation

Motivation: Development of tools for identification of new thioredoxin-fold proteins as well as other proteins belonging to superfamilies with low primary sequence conservation. Results: We present several algorithms for identifying thioredoxin (Trx)-fold proteins containing a conserved CxxC motif (two cysteines separated by two residues). The low conservation of primary sequence in this protein superfamily makes conventional methods difficult to use. Therefore, we use structural properties to build our classifiers. These structural properties include secondary structure patterns as well as various properties of the residues in the protein sequences. We use this information to model Trx-fold proteins via hidden Markov models, decision trees, and algorithms in the multiple-instance learning model. In 9-fold and 12-fold jack-knife tests, some of our models performed quite well, with high true positive and true negative rates. In addition, By combining a small number of our classifiers, we can identify 100 % of the Trx-fold proteins in these jack-knife tests with moderate fals