Validation of Correction of Macromolecular Structural Models

Recent advances in automation in the field of computational structural biology have created a void to be filled by novel validation software. In this project, the problems and inadequacies of currently available validation tools are identified, and the requirements of a novel validation tool are both ascertained and addressed. The development of a new validation software package is described in detail, starting with the development of the front-end interface and the back-end calculations, followed by the integration of these two components to produce an all-in-one validation package, which can calculate its own comprehensive per-residue validation metrics and present them in a compact, interactive, graphical interface, so as to allow the intuitive and thorough analysis of a protein model’s quality that is understandable at a glance. This interface features a novel graphical representation of validation, which plots multiple validation metrics along concentric axes such that correlations between those metrics are immediately apparent, and poorly-modelled regions are emphasised to the user. The software can be run standalone, or plugged into new or existing validation pipelines, and can incorporate calculated metrics from other validation services such as MolProbity (1). It supports multi-model comparison in its single view, and runs with negligible time penalty, making it especially suitable for evaluating incremental changes that result from automated or manual iterative model building. To showcase its extensibility and pluggable design, the integration of this package into the existing CCP4i2 (2) software suite is described. Finally, the package is analysed both quantitatively and qualitatively, and potential avenues for future work are outlined.