A Local Agreement Filtering Algorithm for Transmission EM Reconstructions.

Area of Science:

• Structural Biology
• Biophysics
• Computational Biology

Background:

• Single particle analysis in cryo-electron microscopy (cryo-EM) reconstructs high-resolution molecular volumes from numerous low signal-to-noise images.
• Local variations in signal-to-noise ratio within cryo-EM reconstructions hinder accurate molecular structure interpretation.
• Effective local de-noising is crucial for improving the interpretability and information yield of cryo-EM maps.

Purpose of the Study:

• To introduce LAFTER, a novel local de-noising algorithm designed to enhance single particle reconstructions in cryo-EM.
• To improve the interpretability of molecular structures by addressing local signal-to-noise variations.
• To provide a computationally efficient and freely available tool for the cryo-EM community.

Main Methods:

• LAFTER employs a pair of serial real-space filters to compare independent half-set reconstructions.
• The algorithm identifies and retains shared structural features that exceed noise levels.
• A new evaluation metric for real-space filter effectiveness is proposed, based on Fourier shell correlation (FSC) curve correspondence.

Main Results:

• LAFTER successfully recovers structural features across a broad spectrum of signal-to-noise ratios.
• Demonstrated recovery of strong features at Fourier shell correlation (FSC) values as low as 0.144.
• The algorithm is fast, computationally efficient, and available as a free implementation.

Conclusions:

• LAFTER significantly improves the de-noising of single particle reconstructions in cryo-EM.
• The algorithm enhances the interpretability of molecular structures by preserving genuine features amidst noise.
• The proposed evaluation method offers a robust way to assess real-space filter performance for noise suppression.