Using Combined Features to Analyze Atomic Structures Derived from Cryo-EM Density Maps.

Area of Science:

• Structural Biology
• Biophysics
• Computational Biology

Background:

• Cryo-electron microscopy (cryo-EM) is pivotal for determining protein structures, with atomic detail achievable at high resolutions.
• Accurate side-chain conformation determination is crucial but challenging with standard cryo-EM resolutions (2-4 Å).
• A statistical method is needed to identify anomalous side-chains in cryo-EM models lacking super-resolution data.

Purpose of the Study:

• To develop and validate a statistical method for detecting anomalous side-chains in protein models derived from cryo-EM data.
• To establish a reference dataset using high-resolution X-ray crystallography structures.
• To assess the prevalence of anomalies in cryo-EM models across different resolution ranges and time periods.

Main Methods:

• Analyzed protein structures from X-ray crystallography (<1.5 Å) and cryo-EM (2-4 Å, 4-6 Å) datasets.
• Introduced the histogram-based outlier score (HBOS) for anomaly detection.
• Utilized five features: distal block distance, side-chain length, phi, psi, and first chi angle.

Main Results:

• Cryo-EM models exhibited higher percentages of anomalies compared to high-resolution X-ray models.
• Models derived from cryo-EM data after January 2017 showed fewer anomalies than those from before 2017.
• The HBOS method effectively identified deviations in side-chain conformations.

Conclusions:

• The HBOS method provides a valuable tool for assessing the quality of cryo-EM protein models.
• Improvements in cryo-EM techniques and data processing have led to more accurate side-chain modeling.
• This statistical approach aids in refining protein structures determined by cryo-EM.