Date of Award

6-9-2026

Document Type

Thesis

Publisher

Santa Clara : Santa Clara University, 2026

Departments

Bioengineering; Computer Engineering; Computer Science and Engineering

First Advisor

David Anastasiu

Second Advisor

Hamed Akbari

Abstract

Single-chain fragment variable (scFv) antibodies are powerful, miniature antibodies with success in a variety of cancer therapies. However, improving their thermostability remains a relevant protein design problem, one that we address with interpretable AI-aided design. Machine learning programs have been largely successful in predicting the structure or sequence of proteins, but we still do not know what biophysical characteristics are learned by models to inform these predictions. With the aim of simulating the accuracy of computational physics based models, Message Passing Neural Networks (MPNN) use spatial data to construct feature vector representations of proteins which can then be trained to predict functionality. These models represent proteins as a set of nodes and edges where information about each is first embedded as vectors and is then shared across them via messages. ProteinMPNN uses the 3D coordinates of backbone atoms to build a latent dimension which is then decoded to determine sequences of amino acids likely to fold into the original backbone conformation. We aim to use Sparse Autoencoders (SAEs) to expand the latent dimension and correlate feature dimensions to true protein features to understand what makes ProteinMPNN effective and to steer its predictions. We hope to use these feature dimensions to explain what ProteinMPNN is learning about its input and to imbue output sequences with specific qualities. With this strategy, we successfully increased the theoretical thermostability of Brolucizimab, a commercial scFv drug.

Download

Share

COinS