Neural networks applied to the collagenous disease Osteogenesis imperfecta

Osteogenesis imperfecta is a serious disease which causes bones to be abnormally brittle, and thus, easily broken. It occurs when there is a mutation in the primary sequence of type I collagen. It varies in clinical representation including lethal and non-lethal forms. Severity of the disease appears to be dependent on the type of mutation in the genes COL1A1 and COL1A2. In an attempt to understand the clinical phenotypes of Osteogenesis imperfecta, the authors began an examination of the tertiary structure and primary sequence of collagen type I. The primary sequence of collagen type I was examined to look for neighborhood differences which might lead to specific phenotypes of the disease. Simple patterns were derived and tested. Representation schemes for use with a neural network were found and tested to try to discern the difference between the simplest classification of lethal and non-lethal clinical phenotypes