FluProCAD: a computational screening workflow for fluorescent protein variants

Fluorescent proteins are the backbone of modern high-resolution microscopy, but natural variants often require optimisation to perform effectively. While directed evolution is commonly used for such optimisation, predicting mutation effects requires a deep understanding of photophysics and photochemistry at the atomic scale. Computational chemistry provides a route to obtain such insights from first principles but requires significant expertise. To address this, we developed FluProCAD, a command-line-based workflow that automates system setup and computation of key properties of fluorescent protein mutants using established atomistic models, without the need for prior modeling experience. We applied FluProCAD to two case studies. First, we evaluated the optical and thermodynamic properties of <i>Aequorea victoria</i> Green Fluorescent Protein (<i>av</i>GFP) mutants, successfully reproducing changes in optical responses and folding and dimerisation free energies for five variants. Second, we predicted structural changes in 14 rsGreen0.7 protein variants and validated these models against experimental crystal structures. These results demonstrate the potential of FluProCAD to streamline the optimisation of fluorescent proteins, and expand the computational toolkit for advancing their performance.