Create README.md

README.md

@@ -0,0 +1,3 @@
+# AF-NMR
+
+Recent advances in molecular modeling using deep learning have the potential torevolutionize thefield of structural biology. In particular, AlphaFold has been observedto provide models of protein structures with accuracies rivaling medium-resolution X-raycrystal structures, and with excellent atomic coordinate matches to experimental proteinNMR and cryo-electron microscopy structures. Here we assess the hypothesis thatAlphaFold models of small, relatively rigid proteins have accuracies (based oncomparison  against  experimental  data)  similar  to  experimental  solution  NMRstructures. We selected six representative small proteins with structures determined byboth NMR and X-ray crystallography, and modeled each of them using AlphaFold. Usingseveral structure validation tools integrated under the Protein Structure Validation Softwaresuite (PSVS), we then assessed how well these modelsfit to experimental NMR data,including NOESY peak lists (RPF-DP scores), comparisons between predicted rigidity andchemical shift data (ANSURR scores), and15N-1H residual dipolar coupling data (RDC Qfactors) analyzed by software tools integrated in the PSVS suite. Remarkably, thefits toNMR data for the protein structure models predicted with AlphaFold are generally similar,or better, than for the corresponding experimental NMR or X-ray crystal structures. Similarconclusions were reached in comparing AlphaFold2 predictions and NMR structures forthree targets from the Critical Assessment of Protein Structure Prediction (CASP). Theseresults contradict the widely held misperception that AlphaFold cannot accurately modelsolution NMR structures. They also document the value of PSVS for model vs. dataassessment of protein NMR structures, and the potential for using AlphaFold models forguiding analysis of experimental NMR data and more generally in structural biology.