DeepMainmast and Cryoread: Protein and DNA/RNA Structure Modeling for Cryo-Em Using Deep Learning

In recent years, an increasing number of protein and nucleotide structures have been modeled from cryo-electron microscopy (cryo-EM) maps. Although the resolution of EM maps has improved in general, in practice there are still many cases where modeling biomolecular structures is challenging. To address the needs, we developed structure modeling methods for proteins (DeepMainmast) and nucleic acids (CryoREAD) from cryo-EM maps of a resolution of 2 to 5 Å. Both methods apply deep learning for detecting key atoms and moieties for modeling.