Abstract

The topology of connective tissues, such as tendons, skin, bones or cartilage is controlled to a large extent by the variety of assemblies formed by a specific class of proteins, namely fibrillar collagens. Members of this group make up the most abundant proteins in our body; these polymers and the higher-order structures that they form provide mechanical support within every tissue in which they are present. Although, fundamentally, fibrillar collagen proteins exhibit high level of structural and functional homology, each collagen molecule has its own unique underlying composition, and this composition is responsible for providing the specific designed features of each protein at the molecular level. Given the set of homologous sequences, we build probabilistic models of collagen sequences to identify informative regions and correlations within and among polypeptide chains; revealing which parts of the sequence confer specificity for each collagen chain, making it unique and distinct from others. Furthermore we probe which pairs of sequence positions depend on each other to obtain better understanding of the functionality of collagen molecules.