Home Downloads Tutorials Contact Us Lab

Emap2sec is a computational tool using deep learning that can accurately identify protein secondary structures, alpha helices, beta sheets, others (coils/turns), in cryo-Electron Microscopy (EM) maps of medium to low resolution.

Sai Raghavendra Maddhuri Venkata Subramaniya, Genki Terashi & Daisuke Kihara. Protein Secondary Structure Detection in Intermediate Resolution Cryo-Electron Microscopy Maps Using Deep Learning. Nature Methods (2019).

About Emap2sec

Emap2sec identifies the secondary structures of proteins in cryo-EM maps of intermediate resolution range (~5 to 10 Å) .

Emap2sec uses convolutional deep neural network as its core algorithm and assigns a secondary structure to each of the grid points in an EM map.

Emap2sec architecture has 4 main steps:

(1) Data generation from an input EM map - This step takes in your EM map and formats the map file into an input file for Emap2sec.

(2) Emap2sec phase1 - This is the first phase of our deep learning model consisting of a convolutional neural network (CNN) for local structure detection.

(3) Emap2sec phase2 - This is the second step of our deep learning model and it essentially performs prediction smoothing to eliminate obvious false positives and false negatives.

(4) Voxel-wise secondary structure (SS) assignment - The final step that gives out secondary structure assignments for your EM map.