Primary Immunodeficiency Information Knowledge Services
Acta Universitatis Tamperensis No. 1738
By Crina Samrghitean
July 2012
Tampere University Press
ISBN: 9789514488245
116 pages
$87.50 Paper original
New technologies allow researchers to produce large amounts of data (e.g. from textual and multimedia sources), which represents a challenge for the scientific community. Bioinformatics fills the gaps by creating algorithms, tools and methods to process the increasing quantity of information. The main contribution of this research project was to introduce new and improved biomedical informatics methods in the field of primary immunodeficiency diseases (PIDs). Patients with these diseases have an increased rate of infections but also autoimmune and malignant manifestations. Many of these diseases are very rare with a fatal end and often they are misdiagnosed or have a delayed diagnosis. Developing software systems within the domain of primary immunodeficiencies is a highly challenging task.
In this study two databases were created and a new classification for PIDs was developed. The studies described here use an interdisciplinary approach, based on database and datamining technologies, artificial intelligence, machine learning and combined data from different disciplines such as molecular biology, genetics, immunology, bioinformatics. The wide ranging domain of PIDs was investigated at the protein, genetic, and clinical level and, based on the analyses of different PIDs. Two databases, ImmunoDeficiency Resources (IDR) and IDdiagnostics were developed. IDR is a comprehensive knowledge base for PIDs, which includes tools for clinical, biochemical, genetic, structural and computational analyses as well as links to related information maintained by others. IDdiagnostics is a directory of laboratories performing genetic and clinical tests for PIDs. A concept map for the bioinformatics study of PIDs was designed for different types of users. The model can be used for different types of hereditary diseases.
Several computational methods for the classification and clustering of PIDs have been developed a novel classification of 11 groups, which revealed previously unknown features and relationships of PIDs. These methods aim at automating the classification of PIDs and therefore would be very useful for the PID research and clinical community. Comparison of the classification to independent features such as severity and therapy of the diseases, functional classification of proteins, and network vulnerability, indicated a strong statistical support. The method can be applied to any group of diseases.