Beyond Current Boundaries: Integrating Deep Learning and AlphaFold for Enhanced Protein Structure Prediction from Low-Resolution Cryo-EM Maps

TL;DR

This study introduces DeepTracer-LowResEnhance, an innovative computational framework that uniquely integrates structural predictions from AlphaFold with a deep-learning-based map refinement strategy specifically tailored to enhance low-resolution maps.

Abstract

Constructing atomic models from cryo-electron microscopy (cryo-EM) maps is a crucial yet intricate task in structural biology. While advancements in deep learning, such as convolutional neural networks (CNNs) and graph neural networks (GNNs), have spurred the development of sophisticated map-to-model tools like DeepTracer and ModelAngelo, their efficacy notably diminishes with low-resolution maps beyond 4 Å. To address this shortfall, our research introduces DeepTracer-LowResEnhance, an innovative framework that synergizes a deep learning-enhanced map refinement technique with the power of AlphaFold. This methodology is designed to markedly improve the construction of models from low-resolution cryo-EM maps. DeepTracer-LowResEnhance was rigorously tested on a set of 37 protein cryo-EM maps, with resolutions ranging between 2.5 to 8.4 Å, including 22 maps with resolutions lower than 4 Å. The outcomes were compelling, demonstrating that 95.5\% of the low-resolution maps exhibited a significant uptick in the count of total predicted residues. This denotes a pronounced improvement in atomic model building for low-resolution maps. Additionally, a comparative analysis alongside Phenix's auto-sharpening functionality delineates DeepTracer-LowResEnhance's superior capability in rendering more detailed and precise atomic models, thereby pushing the boundaries of current computational structural biology methodologies.

Figures (15)

• Figure 1: Utilizing high-resolution cryo-EM electron density maps to predict protein structure. Left: cryo-EM density map (EMD-15271). Right: Predicted all-atom structure by running ModelAngelo ModelAngelo.
• Figure 2: Trend of EMDB maps and PDB EM models along the years. (Accessed on May 29th, 2024. emdbpdb)
• Figure 3: DeepTracer pipeline for analyzing cryo-EM maps to predict protein and nucleic acid structures. The process starts with the cryo-EM map input, followed by segmentation using a U-Net convolutional neural network (CNN) and normalization with resizing. A specialized nucleotide U-Net analyzes nucleotides, distinct from the amino acid U-Net. The final output combines the predicted amino acid structures with post-processed nucleotides.
• Figure 4: Workflow of DeepTracer-LowResEnhance. The process begins with the input of a cryo-EM map, optionally accompanied by sequence data. If sequence data is available, it is processed through AlphaFold for an initial 3D structure, refined using ChimeraX, which generates a simulated map. Both maps are then fed into the CryoFEM module, where the simulated map and the cryo-EM map are averaged and split into map chunks. A map reconstructing deep neural network will generate the refined map, which is then processed by DeepTracer to generate a high-accuracy 3D protein structure model.
• Figure 5: A depiction of a limiting scenario with EMD-29072: A 6.0 Å resolution cryo-EM map in gray juxtaposed against its AlphaFold simulated counterpart in brown, highlighting the critical potential for shape mismatch errors.