Protein Sequence Data Research Articles

Machine Learning Techniques to Infer Protein Structure and Function from Sequences: A Comprehensive Review.

The elucidation of protein structure and function plays a pivotal role in understanding biological processes and facilitating drug discovery. With the exponential growth of protein sequence data, machine learning techniques have emerged as powerful tools for predicting protein characteristics from sequences alone. This review provides a comprehensive overview of the importance and application of machine learning in inferring protein structure and function. We discuss various machine learning approaches, primarily focusing on convolutional neural networks and natural language processing, and their utilization in predicting protein secondary and tertiary structures, residue-residue contacts, protein function, and subcellular localization. Furthermore, we highlight the challenges associated with using machine learning techniques in this context, such as the availability of high-quality training datasets and the interpretability of models. We also delve into the latest progress in the field concerning the advancements made in the development of intricate deep learning architectures. Overall, this review underscores the significance of machine learning in advancing our understanding of protein structure and function, and its potential to revolutionize drug discovery and personalized medicine.

Design of ancestral mammalian alkaline phosphatase bearing high stability and productivity.

Sequence-based protein redesign methods, such as full consensus design (FCD) and ancestral sequence reconstruction (ASR), have the potential to construct new enzymes utilizing protein sequence data registered in a rapidly expanding sequence database. The high thermostability of these enzymes would expand their application in diagnostics and molecular biology. These enzymes have also demonstrated a high level of active expression by Escherichia coli. These characteristics make these APs attractive candidates for industrial application. In addition, different amino acid residues are primarily responsible for thermal stability and active expression, suggesting important implications for strategies for designing enzymes by FCD and ASR.

Identifying sequential differences between protein structural classes using network and statistical approaches

Protein sequence information is believed to embed the hint of their structures. To uncover the nature between protein sequence and their structures, this study motivates to inspect the dynamic interactions between various protein sequence features, and identify the sequential differences between the different protein structures. Protein sequence data from all structural classes in CATH and SCOP, and the structural disordered proteins from DisProt, as well as the structural motifs in PROSITE, are analyzed in this study. Betweenness and closeness centrality measures are employed to capture the topology of the networks constructed from amino acid feature interactions, while statistical tests are further implemented to compare the feature series distributions. Key findings suggest that in all structural classes, the features for Ala and α-helix and bend preference property, Ala and side-chain size, Ala and Gly, as well as Met and Leu attain significant interactions between each other, and the feature for Leu, Val, and Asn are acted as the critical sources of feature interactions, whereas Cys, His, Trp, and Met exhibit weak intra-type interactions with other features. These implicate that these feature interactions may have little impact in coding the structural differences. For the α structures, Glu, Pro and side-chain size, hydrophobicity properties exhibit high importance in feature interactions, whereas Gly, Thr and physical properties such as α-helix and bend preference, extended structural preference, pK-C value and surrounding hydrophobicity for β structures, show special high importance in β structures. Both α and β types of structures show Ser as the common sources of feature interactions, while the mixed α and β structures not only show common characters with the α and β types of structures, but also preferred interactions between Met, Lys and double-bend preference property, and between the sequence arrangements of Cys, His, Met, Tyr and amino acid composition features. The intrinsically disordered proteins (IDPs) present high frequency for the repetition patterns of certain amino acids, while the different structural motifs also show special characters. More sequential differences between the structures can also be identified from K-mers statistics and feature series distributions. The new discoveries reveal the nature of amino acid feature interaction mechanics, and show great importance of these interactions in coding the different types of protein structures. The results can not only contribute to future molecular design for protein-based vaccine or drug, but also enlighten the development for new protein structural classifiers.

RecGOBD: accurate recognition of gene ontology related brain development protein functions through multi-feature fusion and attention mechanisms.

Protein function prediction is crucial in bioinformatics, driven by the growth of protein sequence data from high-throughput technologies. Traditional methods are costly and slow, underscoring the need for computational solutions. While deep learning offers powerful tools, many models lack optimization for brain development datasets, critical for neurodevelopmental disorder research. To address this, we developed RecGOBD (Recognition of Gene Ontology-related Brain Development protein function), a model tailored to predict protein functions essential to brain development. RecGOBD targets 10 key gene ontology (GO) terms for brain development, embedding protein sequences associated with these terms. Leveraging advanced pre-trained models, it captures both sequence and structure data, aligning them with GO terms through attention mechanisms. The category attention layer enhances prediction accuracy. RecGOBD surpassed five benchmark models in AUROC, AUPR, and Fmax metrics and was further used to predict autism-related protein functions and assess mutation impacts on GO terms. These findings highlight RecGOBD's potential in advancing protein function prediction for neurodevelopmental disorders. All Python codes associated with this study are available at https://github.com/ZL-Xia/RECGOBD.git.

Identification of Critical Phosphorylation Sites Enhancing Kinase Activity with a Bimodal Fusion Framework

Phosphorylation is an indispensable regulatory mechanism in cells, with specific sites on kinases that can significantly enhance their activity. Although several such critical phosphorylation sites (phos-sites) have been experimentally identified, many more remain to be explored. To date, no computational method exists to systematically identify these critical phos-sites on kinases. In this study, we introduce PhoSiteformer, a transformer-inspired foundational model designed to generate embeddings of phos-sites using phosphorylation mass spectrometry (phos-MS) data. Recognizing the complementary insights offered by protein sequence data and phos-MS data, we developed a classification model, CSPred, which employs a bimodal fusion strategy. CSPred combines embeddings from PhoSiteformer with those from the protein language model ProtT5. Our approach successfully identified 77 critical phos-sites on 58 human kinases. Two of these sites, T517 on PRKG1 and T735 on PRKD3, have been experimentally verified. This study presents the first systematic and computational approach to identify critical phos-sites that enhance kinase activity.

CrossPredGO: A Novel Light-weight Cross-modal Multi-attention Framework for Protein Function Prediction.

Proteins are represented in various ways, each contributing differently to protein-related tasks. Here, information from each representation (protein sequence, 3D structure, and interaction data) is combined for an efficient protein function prediction task. Recently, uni-modal has produced promising results with state-of-the-art attention mechanisms that learn the relative importance of features, whereas multi-modal approaches have produced promising results by simply concatenating obtained features using a computational approach from different representations which leads to an increase in the overall trainable parameters. In this paper, we propose a novel, light-weight cross-modal multi-attention (CrMoMulAtt) mechanism that captures the relative contribution of each modality with a lower number of trainable parameters. The proposed mechanism shows a higher contribution from PPI and a lower contribution from structure data. The results obtained from the proposed CrossPredGO mechanism demonstrate an increment in Fmax in the range of +(3.29 to 7.20)% with at most 31% lower trainable parameters compared with DeepGO and MultiPredGO.

ADTGP: correcting single-cell antibody sequencing data using Gaussian process regression.

We present ADTGP, an R package that uses Gaussian process regression to correct droplet-specific technical noise in single-cell protein sequencing data. ADTGP improves the interpretability of the data by modeling the distribution of protein expression, conditioned on equal isotype control counts across cells. ADTGP is written in R and needs only the protein raw counts, isotype control raw counts, and a design matrix to run. ADTGP can be installed from https://github.com/northNomad/ADTGP. It depends on Stan and the R package 'cmdstanr'.

20 years of the Bio-Analytic Resource for Plant Biology.

The Bio-Analytic Resource for Plant Biology ('the BAR', at https://bar.utoronto.ca) is celebrating its 20th year in operation in 2025. The BAR encompasses and provides visualization tools for large 'omics data sets from plants. The BAR covers data from Arabidopsis, tomato, wheat, barley and 29 other plant species (with data for 2 others to be released soon). These data include nucleotide and protein sequence data, gene expression data, protein-protein and protein-DNA interactions, protein structures, subcellular localizations, and polymorphisms. The data are stored in more than 200 relational databases holding 186 GB of data and are presented to the researchers via web apps. These web apps provide data analysis and visualization tools. Some of the most popular tools are eFP ('electronic fluorescent pictograph') Browsers, ePlantsand ThaleMine (an Arabidopsis-specific instance of InterMine). The BAR was designated a Global Core Biodata Resource in 2023. Like other GCBRs, the BAR has excellent operational stability, provides access without login requirement, and provides an API for researchers to be able to access BAR data programmatically. We present in this update a new overarching search tool called Gaia that permits easy access to all BAR data, powered by machine learning and artificial intelligence.

Protein A-like Peptide Design Based on Diffusion and ESM2 Models

Proteins are the foundation of life, and designing functional proteins remains a key challenge in biotechnology. Before the development of AlphaFold2, the focus of design was primarily on structure-centric approaches such as using the well-known open-source software Rosetta3. Following the development of AlphaFold2, deep-learning techniques for protein design gained prominence. This study proposes a new method to generate functional proteins using the diffusion model and ESM2 protein language model. Diffusion models, which are widely used in image and natural language generation, are used here for protein design, facilitating the controlled generation of new sequences. The ESM2 model, trained on the basis of large-scale protein sequence data, provides a deep understanding of the context of the sequence, thus improving the model’s ability to generate biologically relevant proteins. In this study, we used the Protein A-like peptide as a model study object, combined the diffusion model and the ESM2 model to generate new peptide sequences from minimal input data, and verified their biological activities through experiments such as the BLI affinity test. In conclusion, we developed a new method for protein design that provides a novel strategy to meet the challenges of generic protein generation.

Breaking Down the Bottlebrush: Atomically Detailed Structural Dynamics of Mucins.

Mucins, the biomolecular components of mucus, are glycoproteins that form a thick physical barrier at all tissue-air interfaces, forming a first line of defense against pathogens. Structural features of mucins and their interactions with other biomolecules remain largely unexplored due to the challenges associated with their high-resolution characterization. Combining limited mass spectrometry glycomics and protein sequencing data, we present all-atom, explicitly solvated molecular dynamics simulations of a major respiratory mucin, MUC5B. We detail key forces and degrees of freedom imposed by the extensive O-glycosylation, which imbue the canonically observed bottlebrush-like structures to these otherwise intrinsically disordered protein backbones. We compare our simulation results to static structures observed in recent scanning tunneling microscopy experiments as well as other published experimental efforts. Our work represents the demonstration of a workflow applied to a mucin example, which we hope will be employed by other groups to investigate the dynamics and interactions of other mucins, which can inform on structural details currently inaccessible to experimental techniques.

Impact of phylogeny on the inference of functional sectors from protein sequence data.

Statistical analysis of multiple sequence alignments of homologous proteins has revealed groups of coevolving amino acids called sectors. These groups of amino-acid sites feature collective correlations in their amino-acid usage, and they are associated to functional properties. Modeling showed that nonlinear selection on an additive functional trait of a protein is generically expected to give rise to a functional sector. These modeling results motivated a principled method, called ICOD, which is designed to identify functional sectors, as well as mutational effects, from sequence data. However, a challenge for all methods aiming to identify sectors from multiple sequence alignments is that correlations in amino-acid usage can also arise from the mere fact that homologous sequences share common ancestry, i.e. from phylogeny. Here, we generate controlled synthetic data from a minimal model comprising both phylogeny and functional sectors. We use this data to dissect the impact of phylogeny on sector identification and on mutational effect inference by different methods. We find that ICOD is most robust to phylogeny, but that conservation is also quite robust. Next, we consider natural multiple sequence alignments of protein families for which deep mutational scan experimental data is available. We show that in this natural data, conservation and ICOD best identify sites with strong functional roles, in agreement with our results on synthetic data. Importantly, these two methods have different premises, since they respectively focus on conservation and on correlations. Thus, their joint use can reveal complementary information.

TAWFN: a deep learning framework for protein function prediction.

Proteins play pivotal roles in biological systems, and precise prediction of their functions is indispensable for practical applications. Despite the surge in protein sequence data facilitated by high-throughput techniques, unraveling the exact functionalities of proteins still demands considerable time and resources. Currently, numerous methods rely on protein sequences for prediction, while methods targeting protein structures are scarce, often employing convolutional neural networks (CNN) or graph convolutional networks (GCNs) individually. To address these challenges, our approach starts from protein structures and proposes a method that combines CNN and GCN into a unified framework called the two-model adaptive weight fusion network (TAWFN) for protein function prediction. First, amino acid contact maps and sequences are extracted from the protein structure. Then, the sequence is used to generate one-hot encoded features and deep semantic features. These features, along with the constructed graph, are fed into the adaptive graph convolutional networks (AGCN) module and the multi-layer convolutional neural network (MCNN) module as needed, resulting in preliminary classification outcomes. Finally, the preliminary classification results are inputted into the adaptive weight computation network, where adaptive weights are calculated to fuse the initial predictions from both networks, yielding the final prediction result. To evaluate the effectiveness of our method, experiments were conducted on the PDBset and AFset datasets. For molecular function, biological process, and cellular component tasks, TAWFN achieved area under the precision-recall curve (AUPR) values of 0.718, 0.385, and 0.488 respectively, with corresponding Fmax scores of 0.762, 0.628, and 0.693, and Smin scores of 0.326, 0.483, and 0.454. The experimental results demonstrate that TAWFN exhibits promising performance, outperforming existing methods. The TAWFN source code can be found at: https://github.com/ss0830/TAWFN.

Structure-Based Virtual Screening and Protein–Protein Docking Analysis of ERBB2 and Associated Proteins for Pediatric Cancer Therapeutic Approaches

Abstract Introduction The Erythroblastosis Oncogene B homolog 2 (ERBB2) protein, also known as human epidermal growth factor receptor 2 (HER2), is a key player in cancer growth, especially in neuroblastoma and gastric cancers. Targeting ERBB2 has led to successful therapies, making it an important focus in cancer research with the potential to improve treatment for HER2-positive cancers. Objective The primary goal of this research is to employ a multifaceted computational approach to identify potential drug candidates targeting ERBB2. We aim to combine virtual screening, protein–protein docking, and functional partner prediction to provide insights into the molecular interactions and potential efficacy of the identified compounds. Additionally, we intend to assess the safety profiles of these compounds using advanced toxicity prediction tools. Methods Relevant protein sequence and structural data for ERBB2 and epidermal growth factor receptor (EGFR) were sourced from publicly available databases. Potential inhibitors from the Enamine and LifeChemicals databases were identified through virtual screening using AutoDock Vina. Functional partners of ERBB2 were explored using STRING, KEGG, and REACTOME servers. The identified compounds were subjected to toxicity prediction using the ProTox-II server. Results Virtual screening led to the selection of 10 compounds with favorable binding energies (–8.346 to –6.296 kcal/mol) and specific amino acid interactions (Thr5, Arg412, Leu414, and Ser441) with the receptor. On the other hand, EGFR was identified as the best functional partner for ERBB2. The EGFR residues Gln408, Lys463, Phe412, and Asp436 found key residues for the complex formation. The toxicity prediction analysis revealed that the majority of compounds exhibited acceptable safety profiles, although a subset of compounds showed lower prediction scores, suggesting the need for further consideration. Conclusion This comprehensive computational approach, integrating virtual screening, protein–protein docking, functional partner identification, and toxicity prediction, offers a systematic framework for efficient drug discovery. The identification of potential lead compounds targeting ERBB2, with emphasis on both binding affinity and safety, underscores the significance of such an approach in streamlining the drug development process. By prioritizing compounds with promising efficacy, functional relevance, and acceptable toxicity profiles, this study advances our understanding of potential therapeutic agents, enhancing the likelihood of successful translation from computational predictions to real-world drug candidates.

LearnMSA2: deep protein multiple alignments with large language and hidden Markov models.

For the alignment of large numbers of protein sequences, tools are predominant that decide to align two residues using only simple prior knowledge, e.g. amino acid substitution matrices, and using only part of the available data. The accuracy of state-of-the-art programs declines with decreasing sequence identity and when increasingly large numbers of sequences are aligned. Recently, transformer-based deep-learning models started to harness the vast amount of protein sequence data, resulting in powerful pretrained language models with the main purpose of generating high-dimensional numerical representations, embeddings, for individual sites that agglomerate evolutionary, structural, and biophysical information. We extend the traditional profile hidden Markov model so that it takes as inputs unaligned protein sequences and the corresponding embeddings. We fit the model with gradient descent using our existing differentiable hidden Markov layer. All sequences and their embeddings are jointly aligned to a model of the protein family. We report that our upgraded HMM-based aligner, learnMSA2, combined with the ProtT5-XL protein language model aligns on average almost 6% points more columns correctly than the best amino acid-based competitor and scales well with sequence number. The relative advantage of learnMSA2 over other programs tends to be greater when the sequence identity is lower and when the number of sequences is larger. Our results strengthen the evidence on the rich information contained in protein language models' embeddings and their potential downstream impact on the field of bioinformatics. Availability and implementation: https://github.com/Gaius-Augustus/learnMSA, PyPI and Bioconda, evaluation: https://github.com/felbecker/snakeMSA.

Biological factors associated with long COVID and comparative analysis of SARS-CoV-2 spike protein variants: a retrospective study in Thailand.

Post-acute COVID-19 syndrome (long COVID) refers to the persistence of COVID-19 symptoms or exceptional symptoms following recovery. Even without conferring fatality, it represents a significant global public health burden. Despite many reports on long COVID, the prevalence and data on associated biological factors remain unclear and limited. This research aimed to determine the prevalence of long COVID during the two distinct epidemic periods in Thailand, due to the Delta and Omicron variants of SARS-CoV-2, and to investigate the biological factors associated with long COVID. In addition, the spike protein amino acid sequences of the Delta and Omicron variants were compared to determine the frequency of mutations and their potential biological implications. A retrospective cross-sectional study was established to recruit confirmed COVID-19 participants at Maharat Nakhon Ratchasima Hospital who had recovered for at least three months and were infected between June 2021 and August 2022. The demographic data and long COVID experience were collected via telephone interview. The biological factors were analyzed through binary logistic regression. The datasets of the SARS-CoV-2 spike protein amino acid sequence of the Delta and Omicron variants in Thailand were retrieved from GIDSAID to determine mutation frequencies and to identify possible roles of the mutations based on published data. Data was collected from a total of 247 participants comprising 106 and 141 participants of the Delta and Omicron epidemic periods, respectively. Apart from the COVID-19 severity and health status, the baseline participant data of the two time periods were remarkably similar. The prevalence of long COVID observed in the Omicron period was higher than in the Delta period (74.5% vs. 66.0%). The biological factors associated with long COVID were epidemic variant, age, treatment with symptomatic medicines, and vaccination status. When the spike protein sequence data of the two variants were compared, it was observed that the Omicron variant exhibited a greater quantity of amino acid changes in its receptor-binding domain (RBD) and receptor-binding motif (RBM). The critical changes of the Omicron variant within these regions had a significant function in enhancing virus transmissibility and host immune response resistance. This study revealed informative data associated with long COVID in Thailand. More attention should be given to long COVID caused by unique virus variants and other biological factors to prepare a healthcare management strategy for COVID-19 patients after recovery.

Reassessing the exon–foldon correspondence using frustration analysis

Protein folding and evolution are intimately linked phenomena. Here, we revisit the concept of exons as potential protein folding modules across a set of 38 abundant and conserved protein families. Taking advantage of genomic exon-intron organization and extensive protein sequence data, we explore exon boundary conservation and assess the foldon-like behavior of exons using energy landscape theoretic measurements. We found deviations in the exon size distribution from exponential decay indicating selection in evolution. We show that when taken together there is a pronounced tendency to independent foldability for segments corresponding to the more conserved exons, supporting the idea of exon-foldon correspondence. While 45% of the families follow this general trend when analyzed individually, there are some families for which other stronger functional determinants, such as preserving frustrated active sites, may be acting. We further develop a systematic partitioning of protein domains using exon boundary hotspots, showing that minimal common exons correspond with uninterrupted alpha and/or beta elements for the majority of the families but not for all of them.

ESMSec: Prediction of Secreted Proteins in Human Body Fluids Using Protein Language Models and Attention.

The secreted proteins of human body fluid have the potential to be used as biomarkers for diseases. These biomarkers can be used for early diagnosis and risk prediction of diseases, so the study of secreted proteins of human body fluid has great application value. In recent years, the deep-learning-based transformer language model has transferred from the field of natural language processing (NLP) to the field of proteomics, leading to the development of protein language models (PLMs) for protein sequence representation. Here, we propose a deep learning framework called ESM Predict Secreted Proteins (ESMSec) to predict three types of proteins secreted in human body fluid. The ESMSec is based on the ESM2 model and attention architecture. Specifically, the protein sequence data are firstly put into the ESM2 model to extract the feature information from the last hidden layer, and all the input proteins are encoded into a fixed 1000 × 480 matrix. Secondly, multi-head attention with a fully connected neural network is employed as the classifier to perform binary classification according to whether they are secreted into each body fluid. Our experiment utilized three human body fluids that are important and ubiquitous markers. Experimental results show that ESMSec achieved average accuracy of 0.8486, 0.8358, and 0.8325 on the testing datasets for plasma, cerebrospinal fluid (CSF), and seminal fluid, which on average outperform the state-of-the-art (SOTA) methods. The outstanding performance results of ESMSec demonstrate that the ESM can improve the prediction performance of the model and has great potential to screen the secretion information of human body fluid proteins.

Use of directed quasi-metric distances for quantifying the information of gene families

A large hindrance to analyzing information in genetic or protein sequence data has been a lack of a mathematical framework for doing so. In this paper, we present a multinomial probability space X as a general foundation for multicategory discrete data, where categories refer to variants/alleles of biosequences. The external information that is infused in order to generate a sample of such data is quantified as a distance on X between the prior distribution of data and the empirical distribution of the sample. A number of distances on X are treated. All of them have an information theoretic interpretation, reflecting the information that the sampling mechanism provides about which variants that have a selective advantage and therefore appear more frequently compared to prior expectations. This includes distances on X based on mutual information, conditional mutual information, active information, and functional information. The functional information distance is singled out as particularly useful. It is simple and has intuitive interpretations in terms of 1) a rejection sampling mechanism, where functional entities are retained, whereas non-functional categories are censored, and 2) evolutionary waiting times. The functional information is also a quasi-metric on X, with information being measured in an asymmetric, mountainous landscape. This quasi-metric property is also retained for a robustified version of the functional information distance that allows for mutations in the sampling mechanism. The functional information quasi-metric has been applied with success on bioinformatics data sets, for proteins and sequence alignment of protein families.

Cross-species molecular docking method to support predictions of species susceptibility to chemical effects

The advancement of protein structural prediction tools, exemplified by AlphaFold and Iterative Threading ASSEmbly Refinement, has enabled the prediction of protein structures across species based on available protein sequence and structural data. In this study, we introduce an innovative molecular docking method that capitalizes on this wealth of structural data to enhance predictions of chemical susceptibility across species. We demonstrated this method using the androgen receptor as a pertinent modulator of endocrine function. By using protein structures, this method contextualizes species susceptibility within a functional framework and helps to integrate molecular docking into the repertoire of New Approach Methodologies (NAMs) that support the Next-Generation Risk Assessment (NGRA) paradigm through the novel integration of various open-source tools.

A comprehensive review and comparison of existing computational methods for protein function prediction.

Protein function prediction is critical for understanding the cellular physiological and biochemical processes, and it opens up new possibilities for advancements in fields such as disease research and drug discovery. During the past decades, with the exponential growth of protein sequence data, many computational methods for predicting protein function have been proposed. Therefore, a systematic review and comparison of these methods are necessary. In this study, we divide these methods into four different categories, including sequence-based methods, 3D structure-based methods, PPI network-based methods and hybrid information-based methods. Furthermore, their advantages and disadvantages are discussed, and then their performance is comprehensively evaluated and compared. Finally, we discuss the challenges and opportunities present in this field.