Low-Resolution Interiors & Interfaces Can Achieve High-Resolution Accuracy

Principal Investigator:Jack Snoeyink
Funding Agency:Duke University
Agency Number:159285

Abstract
The current paradigm for building molecular structures, accepted both by the structural biologists and by the biomedical researchers who are the end-users of them, is that high error rates and large uncertainties are unfortunately the inevitable consequence of low resolution. In fact, however, there are constraints, many of which can be phrased in geometric language, that already demonstrate where the errors occur. By uniting the work of the Richardson lab on structure error detection, and the Baker lab on structure prediction, and by developing new algorithms, this team proposes that we can clean up the local inaccuracies in low-resolution (3-4Å) crystal structures of macromolecules so thoroughly that they match well what would be seen at high resolution (1-2Å). The order-of-magnitude improvements in local accuracy we hope to achieve will enable the entire biomedical research community to gain correct and useful biological insights from large, low-resolution structures much more often.