This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. We are developing machine learning methods for evaluating the quality of a predicted protein structure. Essentially, we aim to train a regression algorithm to predict, e.g., the root-mean-squared deviation between a predicted structure and the native protein structure, without knowing the native structure. The core of the project is the development of a wide variety of structural features that will allow our algorithm to discriminate between good and bad structures. For example, we look for proteins that contain large empty pockets on the interior. Such pockets are common in predicted structures but rarely occur in real structures. Given a large set of such features, we train a regression algorithm to predict the quality of previously unseen predicted structures.
Showing the most recent 10 out of 583 publications