Accurate cryo-EM protein particle picking by integrating the foundational AI image segmentation model and specialized U-Net

Abstract

Cryo-electron microscopy (cryo-EM) has revolutionized the field of structural biology by enabling the precise determination of large protein structures. Picking protein particles in cryo-EM micrographs (images) is a crucial step in the cryo-EM-based structure determination. However, existing methods trained on a limited amount of cryo-EM data still cannot accurately pick protein particles from complex, noisy, and heterogenous cryo-EM images. The general foundational artificial intelligence (AI)-based image segmentation model such as the Segment Anything Model (SAM) trained on huge amounts of general image data cannot segment protein particles well because their training data do not include cryo-EM images. In this work, we present a novel approach (CryoSegNet) of integrating the power of the encoder and decoder-based architecture of an attention-gated U-shape network (U-Net) specially designed and trained for cryo-EM particle picking and the SAM. The U-Net is first trained on a large cryo-EM image dataset and then used to generate input from original cryo-EM images for SAM to make particle pickings. CryoSegNet shows both high precision and recall in segmenting protein particles from cryo-EM micrographs, irrespective of protein type, shape, and size. On several independent datasets of various protein types, CryoSegNet outperforms two top machine learning particle pickers crYOLO and Topaz as well as SAM itself. The average resolution of density maps reconstructed from the particles picked by CryoSegNet is 3.05 Å, 15% better than 3.60 Å of Topaz and 49% better than 5.96 Å of crYOLO. Therefore, CryoSegNet can be applied to enhance the resolution of protein structures constructed from both existing and new cryo-EM data.

CryoVirusDB: A Labeled Cryo-EM Image Dataset for AI-Driven Virus Particle Picking

“…Template-based virus particle picking requires experts to pick some initial particles as templates for software tools to search for more particles, which suffers from the presence of ice contamination, radiation damaged particles, carbon areas, and overlapping aggregated particles in micrographs. AI-based particle picking [14] [15] [16] has the best potential to automate the process and overcome the problems of the manual picking and template-based matching, but the development of sophisticated AI-based virus particle picking methods is largely hindered by the lack of high-quality labelled training and test data of virus particles.…”

BlobCUT: A Contrastive Learning Method to Support Small Blob Detection in Medical Imaging

“…Aside from its application in detecting glomeruli in kidney MRI images, with its denoising capabilities and proficiency in detecting small blobs, BlobCUT can extend its utility to various projects in the medical imaging domain, including tasks such as nuclei/cell detection [53,54] and microparticle picking in electron microscopy imaging [55,56]. Taking nuclei detection as an illustrative example: 1.…”

CaLMPhosKAN: Prediction of General Phosphorylation Sites in Proteins via Fusion of Codon-Aware Embeddings with Amino Acid-Aware Embeddings and Wavelet-based Kolmogorov–Arnold Network

“…The dysregulation of phosphorylation, potentially induced by natural toxins or pathogens, can lead to severe diseases such as cancer, Alzheimer's, and heart disease 10 . Consequently, the identification and understanding of phosphorylation sites are critical for developing new therapeutic strategies and insights into drug design [11][12][13] . Phosphorylated proteins can be experimentally identified through methods such as P-labeling and mass spectrometry 14,15 , which enable the labeling of each specific residue within a peptide as either phosphorylated or non-phosphorylated.…”