DMS/NIGMS 1: Challenges in Stochastic Modeling and Computation for Sequential Vaccine Design

Duke University, Durham NC

Investigators

Abstract

Vaccine effectiveness relies on inducing the immune system to generate protective antibodies. Because antibodies are generated by random processes coupled to positive selection, the ability to induce certain rare but desirable antibodies can be limited by the inherent probability of occurrence. This project uses computational modeling to estimate the probability of antibody occurrence, and to infer pathways for the generation of desired antibodies, in order to help design vaccine regimens that maximize the probability of inducing broadly protective antibodies. In order to do so, it addresses several key technical challenges in computational modeling of the stochastic immune system. This project aims to address substantial shortcomings in current B-cell receptor sequence analysis models and software tools by incorporating sequence context-dependent models of somatic hypermutation to develop or improve methods for (1) computing the probability of induction of a mature antibody sequence from its unmutated ancestor, (2) clustering sequences into clones that share evolutionary descent from an unmutated common ancestor, and (3) reconstructing the evolutionary history of the maturation within a B cell clone and inferring the unmutated common ancestor. In each case, the advances will come from the addition of context-dependence to stochastic evolutionary models of maturation, and the development of new efficient approximation algorithms for computation of probabilities under dependent-site models. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.