Deep learning for design of novel spider silk proteins
Abstract
Spider silk fibers have exceptional mechanical properties, which can even outperform materials like Kevlar. This, combined with their sustainable nature makes them interesting for applications in many different fields. The mechanical properties of the fibers are related to the structure of spider silk proteins (spidroins), as well as to the relative abundance of different spidroin types. In this project, we aim to use state-of-the-art and in-house developed deep learning methods to design a large number of new protein sequences that can mimic or improve the spidroins’ structural features. Selected newly designed spidroins will be produced and included in artificially spun silk fibers, which will be tensile tested. This approach will hopefully allow us to both spin fibers with enhanced mechanical properties, and help us clarifying the relationship between the protein structural properties and the mechanical properties of the resulting fibers.
