Automatic and accurate ligand structure determination guided by cryo-electron microscopy maps.

Area of Science:

• Structural Biology
• Biophysics
• Computational Biology

Background:

• Cryo-electron microscopy (cryoEM) and deep learning have advanced protein complex structural studies.
• Accurate determination of ligand conformations remains a challenge in structural biology.

Purpose of the Study:

• To develop EMERALD, an automated tool for determining ligand structures from medium-resolution cryoEM density.
• To assess the robustness and accuracy of EMERALD across various cryoEM datasets.

Main Methods:

• EMERALD utilizes cryoEM density maps to predict ligand and surrounding side-chain conformations.
• The method incorporates a confidence measure for ligand placement.
• The tool was applied to analyze all protein/ligand structures in the EMDB.

Main Results:

• EMERALD accurately predicts ligands in cryoEM maps as low as 4.5 Å resolution.
• Analysis of EMDB structures revealed: 57% matched deposited models, 16% showed alternate conformations, 22% had ambiguous density, and 5% were misplaced.
• High-resolution crystal structures validated alternate conformations found by EMERALD in five cases.

Conclusions:

• EMERALD provides a robust method for automated ligand structure determination in cryoEM.
• The tool's analysis of EMDB data offers insights into ligand modeling accuracy and conformational flexibility.
• EMERALD is critical for advancing the use of cryoEM in solving protein-ligand complexes.