Search for: All records

Abstract This work presents Neural Optimization for Melting-temperature Enabled by Leveraging Translation (NOMELT), a novel approach for designing and ranking high-temperature stable proteins using neural machine translation. The model, trained on over 4 million protein homologous pairs from organisms adapted to different temperatures, demonstrates promising capability in targeting thermal stability. A designed variant of theDrosophila melanogasterEngrailed Homeodomain shows a melting temperature increase of 15.5 K. Furthermore, NOMELT achieves zero-shot predictive capabilities in ranking experimental melting and half-activation temperatures across a number of protein families. It achieves this without requiring extensive homology data or massive training datasets as do existing zero-shot predictors by specifically learning thermophilicity, as opposed to all natural variation. These findings underscore the potential of leveraging organismal growth temperatures in context-dependent design of proteins for enhanced thermal stability. 

ABSTRACT Recent advances in computer hardware and software, particularly the availability of machine learning (ML) libraries, allow the introduction of data-based topics such as ML into the biophysical curriculum for undergraduate and graduate levels. However, there are many practical challenges of teaching ML to advanced level students in biophysics majors, who often do not have a rich computational background. Aiming to overcome such challenges, we present an educational study, including the design of course topics, pedagogic tools, and assessments of student learning, to develop the new methodology to incorporate the basis of ML in an existing biophysical elective course and engage students in exercises to solve problems in an interdisciplinary field. In general, we observed that students had ample curiosity to learn and apply ML algorithms to predict molecular properties. Notably, feedback from the students suggests that care must be taken to ensure student preparations for understanding the data-driven concepts and fundamental coding aspects required for using ML algorithms. This work establishes a framework for future teaching approaches that unite ML and any existing course in the biophysical curriculum, while also pinpointing the critical challenges that educators and students will likely face.