About
Protein crystallization has long been a time- and labor-intensive bottleneck in structure determination. Laboratory automation, including liquid handlers and imagers, has increased experimental throughput and allowed scientists to focus more on experimental design, but at the cost of substantially more image data requiring evaluation. To address this growing burden, Formulatrix integrated MARCO, a scientist-led, Google-developed autoscoring model, into Rock Maker® to automate drop scoring and crystal identification.
While effective, MARCO struggled to identify crystals co-occurring with other classes within the same drop. To overcome this limitation, Formulatrix developed Sherlock, an advanced AI autoscoring model trained on a larger and more diverse dataset. Equipped with new features, including the ability to reliably identify crystals present alongside other classes in a drop, Sherlock offers markedly improved performance over MARCO, establishing it as the autoscoring model of choice for Rock Maker users.