PaStEL: Generator of Pathways with Structural Change on Pseudo Free-Energy Landscape from CryoEM Images

Noisy cryoEM particle images reflect the conformational heterogeneity of biomolecules and have high potential for the study of biological process. As numerous simulation studies have shown, the study of biological process is attributed to the description of the free energy landscapes on the conformational pathways along with collective variable, which is usually difficult to define. In this study, we propose a methodology to automatically generate plausible conformational pathways via the theoretically isometric latent space trained by deep Auto-Encoder model using cryoEM experimental dataset directly. The proposed method of the PaStEL can speedily show structural change on the plausible conformational pathways along with free energy landscape. Solid theoretical guarantees and tests using synthetic cryoEM data have succeeded in obtaining qualitatively correct energy landscapes on the generated plausible pathways. Furthermore, benchmarking with real cryoEM experimental data of 50S Ribosome has successfully demonstrated that the conformational changes with energy landscapes consistent with existing studies without any manual labor. Finally, the PaStEL was applied to spike proteins of SARS-CoV-2 and successfully characterized the difference in the conformational changes between the wild type and the mutant (D614G) focusing on the Receptor Binding Domain regions.