Emap2sec

Emap2sec identifies protein secondary structures in intermediate-resolution cryo-electron microscopy (cryo-EM) maps (5–10 Å) by assigning α-helices, β-sheets, and coils to grid points using a three-dimensional deep convolutional neural network for enhanced structural interpretation.

Key Features:

• Deep Learning Approach: Employs a three-dimensional deep convolutional neural network (3D-CNN) to analyze cryo-EM density and assign secondary structure types (α-helices, β-sheets, coils/turns) to each grid point.
• Resolution Range: Optimized for intermediate-resolution cryo-EM maps, specifically covering 5–10 Å.
• Performance and Validation: Validated on 34 simulated structures at 6.0 and 10.0 Å and 43 experimental maps at 5.0–9.5 Å, demonstrating clear secondary-structure identification and reported improvement over existing methods.

Scientific Applications:

• Interpretation of intermediate-resolution cryo-EM maps: Provides secondary-structure assignments that improve interpretability of 5–10 Å density maps.
• Model building for macromolecular assemblies: Supplies secondary-structure information useful for building and validating models of large complexes.
• Analysis of membrane proteins and challenging targets: Aids structural analysis where high-resolution data are difficult to obtain.

Methodology:

Uses a three-dimensional deep convolutional neural network (3D-CNN) to assign secondary-structure labels to grid points in cryo-EM maps and was evaluated on simulated maps (34 structures at 6.0 and 10.0 Å) and experimental maps (43 maps at 5.0–9.5 Å).