Protein Secondary Structure Detection in Intermediate-Resolution Cryo-EM Maps using Deep Learning

Abstract

Although structures determined at near-atomic resolution are now routinely reported by cryo-electron microscopy (cryo-EM), many density maps are determined at an intermediate resolution, and extracting structure information from these maps is still a challenge. We report a computational method, Emap2sec, that identifies the secondary structures of proteins (α-helices, β-sheets and other structures) in EM maps at resolutions of between 5 and 10 A. Emap2sec uses a three-dimensional deep convolutional neural network to assign secondary structure to each grid point in an EM map. We tested Emap2sec on EM maps simulated from 34 structures at resolutions of 6.0 and 10.0 A, as well as on 43 maps determined experimentally at resolutions of between 5.0 and 9.5 A. Emap2sec was able to clearly identify the secondary structures in many maps tested, and showed substantially better performance than existing methods. Emap2sec uses a deep convolutional neural network to assign protein secondary structures in intermediate-resolution (5–10 A) cryo-EM maps.