Unified Data Resource for 3DEM

摘要

Abstract Three-dimensional transmission electron microscopy (3DEM) has become a cutting-edge method for determining structures of large biological assemblies due to recent technical advances. Hundreds of 3DEM experiments are reported in the literature each year and more than 4400 structures are now available in public archives through our EMDataBank.org website. Since 2014, following the cryo-electron microscopy "resolution revolution," more than 500 maps with resolutions better than 5.0 Å have been released, most with associated coordinate models. It is critically important that 3DEM maps and the models derived from them are of the highest possible quality, and to achieve this aim it is necessary to have community-accepted validation measures. In our current funding period we developed and promoted new methods and infrastructure to determine map accuracy and resolution, as well as novel methods to build and validate map-derived models. In addition, we hosted challenges and meetings to involve the 3DEM and modeling communities in developing new map and model validation standards. We also significantly expanded the underlying representation for 3DEM experimental methods, and collaborated to fully integrate 3DEM into the wwPDB Deposition and Annotation system. We propose here to continue to develop innovative solutions for validating 3DEM-derived structures of macromolecular assemblies, focusing on the moderate to high resolution range, 5-2 Å. To achieve this we will host new challenges formulated to engage the community using increasingly complex data derived by 3DEM methods and provide statistical analysis of their outcomes. In addition, we will develop and evaluate computational protocols that yield quantifiable parameters for validating 3DEM-derived structures, in terms of density map resolvability, map and model correlation and atom position uncertainty of the associated model. We will also improve the validation criteria for nucleic acids, which occur in 30% of all 3DEM derived models. We will use the EMDataBank website to disseminate the results of our investigations and to provide a platform for community engagement and discussions. Our approach will continue to build on the decades of experience of our team in developing tools for 3DEM map generation and modeling, analyzing large assemblies especially those containing nucleic acids, and working with the community to create standards for structure validation.