Protein structure prediction has gained increased attention over the past decades in a wide range of biological disciplines. Creating an accurate visual model of a protein can aid in protein engineering; which has implications in the creation of therapeutic molecules as is the case with antibodies. The third complementarity determining region of the heavy chain of antibodies (CDR-H3) is known to show a large degree of variation in sequence and in length, and therefore has provided difficulties for structure prediction. By separating the CDR-H3 into two logical sections, the apex and base, and using a homology modeling techniques for each section, this study attempts to predict structure for this important region of antibodies. This method also accounts for certain interactions proven to be relevant in CDR-H3 structure to select a suitable parent for modeling an unknown CDR-H3. The selection algorithm was tested using a test set of proteins, selected based on base type, length and diversity. Overall, there seemed to be a slight improvement in the prediction of CDR-H3 by this method when compared with traditional homology methods; although both drastic improvements and evident decreases in accuracy of predictions from individual molecules can be observed.
Library of Congress Subject Headings
Immunoglobulins--Models; Monoclonal antibodies--Models
Department, Program, or Center
Biomedical Sciences (CHST)
Skuse, Gary - Chair
Galens, Kevin, "Knowledge based structure modeling of the third hypervariable region of antibodies" (2006). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus