Bioinformatics: Development of automated protein family classification using hidden Markov models for functional characterization of proteins

There is a great need to subdivide large protein families into smaller, homogeneous subfamilies, corresponding to functional entities. Hidden Markov models constitute a powerful technique for such subclassification.

This project aims at development of automated strategies, making it possible to utilise these methods on large scale analyses, e.g. complete genomes. The developed automated scheme has been applied on the large MDR superfamily of medium-chain dehydrogenases/reductases, enabling the identification of number of interesting human families. We have also found correlation between zinc content and dehydrogenase activity versus absence of zinc and reductase activity. More information is available at

External link:


Joel Hedlund