Journal
PROTEIN ENGINEERING DESIGN & SELECTION
Volume 23, Issue 9, Pages 689-697Publisher
OXFORD UNIV PRESS
DOI: 10.1093/protein/gzq043
Keywords
antibody modelling; antibody structure; feature selection; humanization; machine learning
Funding
- BBSRC Dorothy Hodgkin Postgraduate Award
Ask authors/readers for more resources
The packing of V-H and V-L domains in antibodies can vary, influencing the topography of the antigen-combining site. However, until recently, this has largely been ignored in modelling antibody structure. We present an analysis of the degree of variability observed in known structures together with a machine-learning approach to predict the packing angle. A neural network was trained on sets of interface residues and a genetic algorithm designed to perform 'feature selection' to define which sets of interface residues could be used most successfully to perform the prediction. While this training procedure was very computationally intensive, prediction is performed in a matter of seconds. Thus, not only do we provide a rapid method for predicting the packing angle, but also we define a set of residues that may be important in antibody humanization in order to obtain the correct binding site topography.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available