Biology subjects

Si, D.

Publications and source records attributed to Si, D..

At least 19 recordsLinked to original sources

bioRxiv2026

Towards building a World Model to simulate perturbation-induced cellular dynamics by AlphaCell

Predicting cellular responses to perturbations is crucial for therapeutic discovery, yet experimental screening is severely constrained by the combinatorial vastness of biological space. While computational simulations offer a scalable alternative, current models are limited by the incomplete latent representation----mainly relying on highly variable genes in feature representation; the poor genome-wide reconstruction fidelity; and the ungeneralizable dynamic laws across diverse contexts. Consequently, they fail to mechanistically transfer learned dynamics to unseen cellular contexts. To address these systemic flaws, we introduce AlphaCell, a generative Virtual Cell World Model that unifies genome-wise representation with continuous state transition modeling. AlphaCell achieves three synergistic innovations: (1) Latent Manifold Rectification, processing the full protein-coding transcriptome to construct a differentiable Virtual Cell Space, effectively filtering noise while preserving intrinsic cellular topology; (2) Biological Reality Reconstruction, utilizing a massive, knowledge-rich decoder to translate abstract latent states back into high-fidelity, genome-wide expression profiles; and (3) Universal State Transition, applying Optimal Transport Conditional Flow Matching to model perturbations as continuous, deterministic vector fields. By abstracting perturbation mechanisms into generalized dynamic laws, AlphaCell makes robust prediction of perturbation responses in a compositional generalization scenario and enables zero-shot prediction of cellular dynamics in entirely unseen cellular contexts, providing a foundational engine for cellular-context-generalizable perturbation prediction and perturbation-induced cellular dynamics simulation.

bioinformatics↗

bioRxiv2026

A Large-Scale Cryo-EM RNA Motif Dataset and Benchmark for Machine Learning-Based Structure Modeling

MotivationRNA molecules play critical roles in gene regulation, viral replication, and cellular control, with their functions tightly coupled to three-dimensional structure. Advances in cryogenic electron microscopy (cryo-EM) now enable RNA structure characterization across a broad resolution range. RNA secondary structural motifs, including hairpins, internal loops, and bulges, act as fundamental building blocks of RNA tertiary architecture and are key targets in RNA-focused therapeutic design. Despite this, most computational approaches for RNA structure prediction from cryo-EM density maps do not explicitly utilize secondary structural motifs as intermediate representations, largely due to the absence of large-scale, high-quality, and motif-resolved datasets suitable for machine learning. ResultsHere, we present a large, open-source dataset containing over 125,000 motif-resolved cryo-EM density maps paired with corresponding atomic structures, spanning 25 classes of RNA secondary structural motifs. The dataset covers resolutions from 1.5 [A] to 34.0 [A], encompassing both near-atomic and low-resolution density maps relevant to RNA modeling. Each motif instance includes a segmented cryo-EM density map represented as a standardized 3D voxel grid, with atomic-level motif annotations propagated to voxel-level labels for RNA backbone, ribose sugar, and nucleobase components. Segmentation quality is validated via cross-correlation analysis, demonstrating strong agreement between motif-level density maps and atomic reference models. To illustrate the datasets utility, high-resolution maps (1.5-2.8 [A]) were used to train a machine learning classifier that distinguished five motif classes with a specificity of 0.948. Availability and ImplementationSource code, implementation of the fully automated pipeline, and the benchmark datasets are publicly available at GitHubhttps://github.com/DrDongSi/3DEM-RNA-Motif-Dataset Zenodohttps://zenodo.org/communities/3dem-rna-motif-dataset Contacthoujie@msu.edu, dongsi@uw.edu

bioinformatics↗

bioRxiv2025

ContrastQA: A Label-guided Graph Contrastive Learning-based approach for protein complex structure quality assessment

Despite recent progress, the Estimation of Model Accuracy (EMA) for protein complexes remains less advanced compared to that for protein monomers. A key challenge lies in effectively integrating both interface-specific and global structural information to accurately assess the quality of protein complexes. Here, we introduce ContrastQA, the first EMA framework for protein complexes that incorporates the proposed label-guided graph contrastive learning based on interface quality. By integrating a geometric graph neural network to model global structural features, ContrastQA effectively captures both local (interface-level) and global (structure-level) information for accurate model quality estimation. ContrastQA achieved ranking losses of 0.123 and 0.116 on the TMscore and GDT-TS metrics on the CASP16 dataset, which are 0.015 (10.9%) and 0.012 (8.7%) lower than the second-best EMA method with ranking losses of 0.138 and 0.128. Furthermore, for the more challenging CASP16 hard targets, our method achieved a TMscore ranking loss of 0.131, which is 0.023 (14.9%) lower than that of the second-best EMA method with a ranking loss of 0.154. Our study demonstrates the strong effectiveness of the label-guided graph contrastive learning module. These findings suggest that our graph contrastive learning framework serves as a valuable pre-training strategy for learning protein structure representations. The source code is available at https://github.com/Cao-Labs/ContrastQA. Significance StatementHigh-quality proteins generated by high-precision structural prediction methods for protein complexes helping biologists prioritize models for drug discovery and understanding disease mechanisms. Recent advances in predicting protein complex structures (like AlphaFold3) still struggle to match experimental precision. A critical bottleneck lies in accurately estimating the quality of predicted models-known as Estimation of Model Accuracy (EMA). Here, we introduce ContrastQA, implementing label-guided graph contrastive learning framework to explore the generalization of high-quality proteins. Our method demonstrates outstanding performance, provides an effective solution for assessing the structural quality of protein complexes.

bioinformatics↗

bioRxiv2025

Benchmarking single-cell multi-modal data integrations

Single-cell multi-modal omics, remarked as "Method of the Year 2019" by Nature Methods, provides valuable insights into complex tissues. Recent advances have enabled the generation both unpaired (separate profiling) and paired (simultaneous measurement) multi-modal datasets, driving rapid development of single-cell multi-modal integration tools. Nevertheless, there is a pressing need for a comprehensive benchmark to assess algorithms under varying integrated dataset types, integrated modalities, dataset sizes, and data quality. Here, we present a systematic benchmark for 40 single-cell multi-modal integration algorithms involving modalities of DNA, RNA, protein and spatial omics for paired, unpaired and mosaic datasets (a mixture of paired and unpaired datasets). We evaluated usability, accuracy, and robustness to assist researchers in selecting suitable integration methods tailored to their datasets and applications. Our benchmark provides valuable guidance in the ever-evolving field of single-cell multi-omics. The Stage 1 protocol for this Registered Report was accepted in principle on 3rd Jul 2024. The protocol, as accepted by the journal, can be found at https://springernature.figshare.com/articles/journal_contribution/Benchmarking_single-cell_multi-modal_data_integrations/26789572.

bioinformatics↗

bioRxiv2024

GOBoost: Leveraging Long-Tail Gene Ontology Terms for Accurate Protein Function Prediction

MotivationWith the advancement of deep learning, researchers have increasingly proposed computational methods based on deep learning techniques to predict protein function. However, many of these methods treat protein function prediction as a multi-label classification problem, often overlooking the long-tail distribution of functional labels (i.e., Gene Ontology Terms) in datasets. To address this issue, we propose the GOBoost method, which incorporates the proposed long-tail optimization ensemble strategy. Besides, GOBoost introduces the proposed global-local label graph module and multi-granularity focal loss function to enhance long-tail functional information, mitigate the long-tail phenomenon, and improve overall prediction accuracy. ResultsWe evaluate GOBoost and other state-of-the-art (SOTA) protein function prediction methods on the PDB and AF2 datasets. The GOBoost outperformed SOTA methods across all evaluation metrics on both datasets. Notably, in the AUPR evaluation on the PDB test set, GOBoost improved by 10.71%, 35.91%, and 22.71% compared to the SOTA HEAL method in the MF, BP, and CC functions. The experimental results demonstrate the necessity and superiority of designing models from the label long-tail distribution perspective. Availabilityhttps://github.com/Cao-Labs/GOBoost Contactcaora@plu.edu

bioinformatics↗

bioRxiv2024

Gradual labeling with fluorogenic probes for MINFLUX nanoscopy of densely packed structures

Single-molecule localization in densely packed structures represents a formidable challenge in nanoscopy. Here we develop a gradual labeling method with fluorogenic probes for MINFLUX nanoscopy (GLF-MINFLUX), with which individual proteins in packed structures can be sequential illuminated and quantitatively localized at the nanoscale solely by adjusting probe concentration. With a 1.7-fold localization precision improvement and 2.2-fold faster acquisition, GLF-MINFLUX enables dual-channel nanoscale imaging of densely neuronal microtubules and microfilaments, and three-dimensional quantitative mapping of individual TOM20 proteins in mitochondrial clusters.

biophysics↗

bioRxiv2024

PerturBase: a comprehensive database for single-cell perturbation data analysis and visualization

Single-cell perturbation sequencing techniques (scPerturbation), represented by single cell genetic perturbation sequencing (e.g., Perturb-seq) and single cell chemical perturbation sequencing (e.g., sci-Plex), result from the integration of single-cell toolkits with conventional bulk screening methods. These innovative sequencing techniques empower researchers to dissect perturbation functions and mechanisms in complex biological systems at an unprecedented resolution. Despite these advancements, a notable gap exists in the availability of a dedicated database for exploring and querying scPerturbation data. To address this gap and facilitate seamless data sharing for researchers, we present PerturBase--the first and most comprehensive database designed for the analysis and visualization of scPerturbation data (http://www.perturbase.cn/). PerturBase consolidates 122 datasets from 46 publicly accessible research studies, covering 115 single-modal and 7 multi-modal datasets that include 24254 genetic and 230 chemical perturbations from about 6 million cells. The database provides insights through various software-analyzed results, encompassing quality control, denoising, differential expression gene analysis, perturbation function analysis, and correlation characterization between perturbations. All datasets and in-depth analyses are presented in user-friendly, easy-to-browse pages and can be visualized through intuitive tables and various image formats. In summary, PerturBase stands as a pioneering high-content database, intended for searching, visualizing, and analyzing scPerturbation datasets, contributing to an enhanced understanding of perturbation functions and mechanisms. Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=159 SRC="FIGDIR/small/578767v1_ufig1.gif" ALT="Figure 1"> View larger version (56K): org.highwire.dtl.DTLVardef@15bdfcborg.highwire.dtl.DTLVardef@c1be77org.highwire.dtl.DTLVardef@1bf7e32org.highwire.dtl.DTLVardef@f62e8b_HPS_FORMAT_FIGEXP M_FIG C_FIG

cell biology↗

bioRxiv2023

Protein-Ligand Binding Site Prediction and de Novo Ligand Generation from Cryo-EM Maps

Identification of protein-ligand binding sites is one of the most challenging tasks in drug discovery and design. Recent advances in machine learning community, particularly deep learning, inspired considerable research into deep learning-based methods for protein ligand binding site prediction (PLBP) and have achieved promising results. However one limitation of these methods and models is that they are developed and evaluated using conventional databases consisting of relatively small protein structures determined from X-ray crystallography and NMR with high resolution. When current PLBP methods are directly applied to large protein complexes, they either fail or produce very poor results. Given the increasing popularity of protein structures determined from Cryo-EM maps, which usually capture large protein complexes at relatively low resolution, there is a strong need to apply PLBP methods to Cryo-EM maps. For this purpose, we created a novel database (EMD-Ligand Dataset) consisting of Cryo-EM maps and information about their ligand-binding partners. Our study of several state-of-the-art PLBP methods show that even though they perform reasonably well on conventional databases, they do poorly on the novel EMD-Ligand Dataset, which calls for more research and work to be done by the community. As current methods for determining protein structures experimentally from Cryo-EM maps are time-consuming and expensive, we integrated PLBP methods with DeepTracer, an automatic and fast, de novo Cryo-EM protein structure modeling method, and established an end-to-end system capable of predicting ligand binding sites directly from Cryo-EM maps. Additionally, we leveraged an existing ligand generator to generate drug-like ligands from our predicted ligand-binding pockets and demonstrated its effectiveness.

bioinformatics↗

bioRxiv2023

Protein Structure Refinement via DeepTracer and AlphaFold2

Understanding the structures of proteins has numerous applications, such as vaccine development. It is a slow and labor-intensive task to manually build protein structures from experimental electron density maps, therefore, machine learning approaches have been proposed to automate this process. However, most of the experimental maps are not atomic resolution, so they are insufficient for computer vision-based machine learning methods to precisely determine the protein structure. On the other hand, methods that utilize evolutionary information from protein sequences to predict structures, like AlphaFold2, have recently achieved groundbreaking accuracy but often require manual effort to refine the results. We propose DeepTracer-Refine, an automated method to refine AlphaFold structures by aligning them to DeepTracers predicted structure. We tested our method on 39 multi-domain proteins and we improved the average residue coverage from 78.2% to 90.0% and average lDDT score from 0.67 to 0.71. We also compared DeepTracer-Refine against another method, Phenixs AlphaFold refinement, to demonstrate that our method not only performs better when the initial AlphaFold model is less precise but also exceeds Phenix in run-time performance.

bioinformatics↗

bioRxiv2023

AnglesRefine: refinement of 3D protein structures using Transformer based on torsion angles

MotivationThe goal of protein structure refinement is to enhance the precision of predicted protein models, particularly at the residue level of the local structure. Existing refinement approaches primarily rely on physics, whereas molecular simulation methods are resource-intensive and time-consuming. In this study, we employ deep learning methods to extract structural constraints from protein structure residues to assist in protein structure refinement. We introduce a novel method, AnglesRefine, which focuses on a proteins secondary structure and employs a transformer model to refine various protein structure angles (psi, phi, omega, CA_C_N_angle, C_N_CA_angle, N_CA_C_angle), ultimately generating a superior protein model based on the refined angles. ResultsWe evaluate our approach against other cutting-edge protein structure refinement methods using the CASP11-14 and CASP15 datasets. Experimental outcomes indicate that our method generally surpasses other techniques on the CASP11-14 test dataset, while performing comparably or marginally better on the CASP15 test dataset. Our method consistently demonstrates the least likelihood of model quality degradation, e.g., the degradation percentage of our method is less than 10%, while other methods are about 50%. Furthermore, as our approach eliminates the need for conformational search and sampling, it significantly reduces computational time compared to existing protein structure refinement methods. Availabilityhttps://github.com/Cao-Labs/AnglesRefine.git Contactcaora@plu.edu CCS CONCEPTS*Computing methodologies [->] Machine learning. ACM Reference FormatLei Zhang, Junyong Zhu, Sheng Wang, Jie Hou, Dong Si, and Renzhi Cao. 2023. AnglesRefine: refinement of 3D protein structures using Transformer based on torsion angles. In Proceedings of 22nd International Workshop on Data Mining in Bioinformatics (BIOKDD 2023) (BIOKDD 2023). ACM, New York, NY, USA, 10 pages. https://doi.org/XXXXXXX.XXXXXXX

bioinformatics↗

bioRxiv2022

Fast and Automated Protein-DNA/RNA Macromolecular Complex Modeling from Cryo-EM Maps

Cryo-electron microscopy (cryo-EM) allows a macromolecular structure such as protein-DNA/RNA complexes to be reconstructed in a three-dimensional coulomb potential map. The structural information of these macromolecular complexes forms the foundation for understanding the molecular mechanism including many human diseases. However, the model building of large macromolecular complexes is often difficult and time-consuming. We recently developed DeepTracer-2.0, an artificial intelligence-based pipeline that can build amino acid and nucleic acid backbones from a single cryo-EM map, and even predict the best-fit residues according to the density of side chains. The experiments showed improved accuracy and efficiency when benchmarking the performance on independent experimental maps of protein-DNA/RNA complexes and demonstrated the promising future of macromolecular modeling from cryo-EM maps. Our method and pipeline could benefit researchers worldwide who work in molecular biomedicine and drug discovery, and substantially increase the throughput of the cryo-EM model building. The pipeline has been integrated into the web portal https://deeptracer.uw.edu/.

bioinformatics↗

bioRxiv2022

Characterisation of an Escherichia coli line that completely lacks ribonucleotide reduction yields insights into the evolution of obligate intracellularity.

All life requires ribonucleotide reduction for de novo synthesis of deoxyribonucleotides. A handful of obligate intracellular species are known to lack ribonucleotide reduction and are instead dependent on their host for deoxyribonucleotide synthesis. As ribonucleotide reduction has on occasion been lost in obligate intracellular parasites and endosymbionts, we reasoned that it should in principle be possible to knock this process out entirely under conditions where deoxyribonucleotides are present in the growth media. We report here the creation of a strain of E. coli where all three ribonucleotide reductase operons have been fully deleted. Our strain is able to grow in the presence of deoxyribonucleosides and shows slowed but substantial growth. Under limiting deoxyribonucleoside levels, we observe a distinctive filamentous cell morphology, where cells grow but do not appear to divide regularly. Finally, we examined whether our lines are able to adapt to limited supplies of deoxyribonucleosides, as might occur in the evolutionary switch from de novo synthesis to dependence on host production during the evolution of parasitism or endosymbiosis. Over the course of an evolution experiment, we observe a 25-fold reduction in the minimum concentration of exogenous deoxyribonucleosides necessary for growth. Genome analysis of replicate lines reveals that several lines carry mutations in deoB and cdd. deoB codes for phosphopentomutase, a key part of the deoxyriboaldolase pathway, which has been hypothesised as an alternative to ribonucleotide reduction for deoxyribonucleotide synthesis. Rather than synthesis via this pathway complementing the loss of ribonucleotide reduction, our experiments reveal that mutations appear that reduce or eliminate the capacity for this pathway to catabolise deoxyribonucleotides, thus preventing their loss via central metabolism. Mutational inactivation of both deoB and cdd is also observed in a number of obligate intracellular bacteria that have lost ribonucleotide reduction. We conclude that our experiments recapitulate key evolutionary steps in the adaptation to intracellular life without ribonucleotide reduction.

microbiology↗

bioRxiv2022

Structure of Geobacter OmcZ filaments suggests extracellular cytochrome polymers evolved independently multiple times

While early genetic and low resolution structural observations suggested that extracellular conductive filaments on metal reducing organisms such as Geobacter were composed of Type IV pili, it has now been established that bacterial c-type cytochromes can polymerize to form extracellular filaments capable of long-range electron transport. Atomic structures exist for two such cytochrome filaments, formed from the hexaheme cytochrome OmcS and the tetraheme cytochrome OmcE. Due to the highly conserved heme packing within the central OmcS and OmcE cores, and shared pattern of heme coordination between subunits, it has been suggested that these polymers have a common origin. We have now used cryo-EM to determine the structure of a third extracellular filament, formed from the Geobacter sulfurreducens octaheme cytochrome, OmcZ. In contrast to the linear heme chains in OmcS and OmcE, the packing of hemes, heme:heme angles, and between-subunit heme coordination is quite different in OmcZ. A branched heme arrangement within OmcZ leads to a highly surface exposed heme in every subunit, which may account for the formation of conductive biofilm networks, and explain the higher measured conductivity of OmcZ filaments. This new structural evidence suggests that conductive cytochrome polymers arose independently on more than one occasion from different ancestral multiheme proteins.

biophysics↗

bioRxiv2022

DeepTracer ID: De Novo Protein Identification from Cryo-EM Maps

The recent revolution in cryo-electron microscopy has made it possible to determine macromolecular structures directly from cell extracts. However, identifying the correct protein from the cryo-EM map is still challenging and often needs additional sequence information from other techniques, such as tandem mass spectrometry and/or bioinformatics. Here, we present DeepTracer-ID, a server-based approach to identify the candidate protein in a user-provided organism de novo from a cryo-EM map, without the need for additional information. Our method first uses DeepTracer to generate a protein backbone model that best represents the cryo-EM map, and this model is then searched against the library of AlphaFold2 predictions for all proteins in the given organism. This method is highly accurate and robust: in all 13 experimental maps tested blindly, DeepTracer-ID identified the correct proteins as the top candidates. Eight of the maps were of known structures, while the other five unpublished maps were validated by prior protein annotation and careful inspection of the model refined into the map. The program also showed promising results for both homomeric and heteromeric protein complexes. This platform is possible because of the recent breakthroughs in large-scale protein 3D structure prediction. Statement of SignificanceWhile it has now become routine for cryo-EM maps of proteins to reach a near-atomic resolution, potentially allowing for reliable atomic models to be built, there are a growing number of instances where the protein identity may not be known. Without knowing the protein sequence, it is impossible to build an atomic model. DeepTracer-ID is a server-based approach to surmount this problem by identifying the proteins in a given organism that are found in the cryo-EM map. A free web service for global academic access is provided.

biophysics↗

bioRxiv2021

Structural Insights Into the Initiation and Elongation of Ubiquitination by Ubr1

The N-end rule pathway was one of the first ubiquitin (Ub)-dependent degradation pathways to be identified. Ubr1, a single-chain E3 ligase, targets proteins bearing a destabilizing residue at the N-terminus (N-degron) for rapid K48-linked ubiquitination and proteasome-dependent degradation. How Ubr1 catalyses the initiation of ubiquitination on the substrate and elongation of the Ub chain in a linkage-specific manner through a single E2 ubiquitin-conjugating enzyme (Ubc2) remains unknown. Here, we report the cryo-electron microscopy structures of two complexes representing the initiation and elongation intermediates of Ubr1 captured using chemical approaches. In these two structures, Ubr1 adopts different conformations to facilitate the transfer of Ub from Ubc2 to either an N-degron peptide or a monoubiquitinated degron. These structures not only reveal the architecture of the Ubr1 complex but also provide mechanistic insights into the initiation and elongation steps of ubiquitination catalysed by Ubr1.

biophysics↗

bioRxiv2021

SynthQA - Hierarchical Machine Learning-based Protein Quality Assessment

MotivationIt has been a challenge for biologists to determine 3D shapes of proteins from a linear chain of amino acids and understand how proteins carry out lifes tasks. Experimental techniques, such as X-ray crystallography or Nuclear Magnetic Resonance, are time-consuming. This highlights the importance of computational methods for protein structure predictions. In the field of protein structure prediction, ranking the predicted protein decoys and selecting the one closest to the native structure is known as protein model quality assessment (QA), or accuracy estimation problem. Traditional QA methods dont consider different types of features from the protein decoy, lack various features for training machine learning models, and dont consider the relationship between features. In this research, we used multi-scale features from energy score to topology of the protein structure, and proposed a hierarchical architecture for training machine learning models to tackle the QA problem. ResultsWe introduce a new single-model QA method that incorporates multi-scale features from protein structures, utilizes the hierarchical architecture of training machine learning models, and predicts the quality of any protein decoy. Based on our experiment, the new hierarchical architecture is more accurate compared to traditional machine learning-based methods. It also considers the relationship between features and generates additional features so machine learning models can be trained more accurately. We trained our new tool, SynthQA, on the CASP dataset (CASP10 to CASP12), and validated our method on 33 targets from the latest CASP 14 dataset. The result shows that our method is comparable to other state-of-the-art single-model QA methods, and consistently outperforms each of the 14 used features. Availabilityhttps://github.com/Cao-Labs/SynthQA.git Contactcaora@plu.edu

bioinformatics↗

bioRxiv2020

DeepTracer: Automated Protein Complex Structure Prediction from CoV-related Cryo-EM Density Maps

Information about macromolecular structure of protein complexes such as SARS-CoV-2, and related cellular and molecular mechanisms can assist the search for vaccines and drug development processes. To obtain such structural information, we present DeepTracer, a fully automatic deep learning-based method for fast de novo multi-chain protein complex structure determination from high-resolution cryo-electron microscopy (cryo-EM) density maps. We applied DeepTracer on a previously published set of 476 raw experimental density maps and compared the results with a current state of the art method. The residue coverage increased by over 30% using DeepTracer and the RMSD value improved from 1.29[A] to 1.18[A]. Additionally, we applied DeepTracer on a set of 62 coronavirus-related density maps, among them 10 with no deposited structure available in EMDataResource. We observed an average residue match of 84% with the deposited structures and an average RMSD of 0.93[A]. Additional tests with related methods further exemplify DeepTracers competitive accuracy and efficiency of structure modeling. DeepTracer allows for exceptionally fast computations, making it possible to trace around 60,000 residues in 350 chains within only two hours. The web service is globally accessible at https://deeptracer.uw.edu.

bioinformatics↗

bioRxiv2020

Outcomes of the 2019 EMDataResource model challenge: validation of cryo-EM models at near-atomic resolution

This paper describes outcomes of the 2019 Cryo-EM Map-based Model Metrics Challenge sponsored by EMDataResource (www.emdataresource.org). The goals of this challenge were (1) to assess the quality of models that can be produced using current modeling software, (2) to check the reproducibility of modeling results from different software developers and users, and (3) compare the performance of current metrics used for evaluation of models. The focus was on near-atomic resolution maps with an innovative twist: three of four target maps formed a resolution series (1.8 to 3.1 [A]) from the same specimen and imaging experiment. Tools developed in previous challenges were expanded for managing, visualizing and analyzing the 63 submitted coordinate models, and several novel metrics were introduced. The results permit specific recommendations to be made about validating near-atomic cryo-EM structures both in the context of individual laboratory experiments and holdings of structure data archives such as the Protein Data Bank. Our findings demonstrate the relatively high accuracy and reproducibility of cryo-EM models derived from these benchmark maps by 13 participating teams, representing both widely used and novel modeling approaches. We also evaluate the pros and cons of the commonly used metrics to assess model quality and recommend the adoption of multiple scoring parameters to provide full and objective annotation and assessment of the model, reflective of the observed density in the cryo-EM map.

biochemistry↗