Matthew Chan is a fourth year gradate student in our group working on understanding mechanisms of neurotransmitter transport and developing software to predict the effect of mutations on protein function. As a MolSSI software fellow, Matthew will help develop a software to enable transfer learning of mutations using deep mutational scanning datasets. As the function of a protein is intrinsically linked to its sequence and structure, predicting the functional effects of point mutations remains a grand challenge among biochemists, biophysicists, and computational biologists. Biochemical workflows such as deep mutational scanning (DMS) leverages high throughput techniques to access the functional consequence of all single point mutations in a sequence1. Computational approaches including MD simulations provide an invaluable perspective of conformational dynamics at atomistic resolution. These two approaches have been used to provide a glimpse into the elusive sequence-structure- function relationship of proteins, but at high experimental cost. One approach to overcome this data scarcity is to utilize transfer learning of the data obtain from DMS and MD simulations of one protein and apply it to homologous proteins within its family. Traditionally, machine learning approaches have been used to enhance predictions in the low-data regime, however, in this case, transfer learning takes advantage of the structural and functional similarities between homologous proteins in order to provide new insights without the high experimental cost.