About
Identifying protein crystals in high-throughput crystallization campaigns remains a persistent bottleneck, as protein and non-protein crystals appear identical under brightfield imaging and often require X-ray analysis for confirmation. At the same time, advances in automation have dramatically increased the volume of images, making manual crystal identification a major challenge.
UV fluorescence imaging addresses the challenge by selectively visualizing protein crystals by utilizing the intrinsic fluorescence of tryptophan. Formulatrix has exploited the full potential of this modality at scale by developing Sherlock UV, an advanced AI-based autoscoring model integrated within Rock Maker®. Trained on a large, diverse dataset of real-world crystallization experiments, Sherlock UV leverages deep learning to reliably distinguish protein crystals from non-protein material and further classify based on their morphology. This application note demonstrates that combining UV fluorescence imaging and AI-based scoring yields a more accurate, scalable, and robust solution, accelerating the path from crystal hit to X-ray diffraction data.