Deep learning for mining protein data

NeuroScale: evolutional scale-based protein language models enable prediction of neuropeptides

BACKGROUND

Neuropeptides (NPs) are critical signaling molecules involved in various physiological and behavioral processes, including development, metabolism, and memory. They function within both the nervous and endocrine systems and have emerged as promising therapeutic targets for a range of diseases. Despite their significance, the accurate identification of NPs remains a challenge, necessitating the development of more effective computational approaches.

RESULTS

In this study, we introduce NeuroScale, a multi-channel neural network model leveraging evolutionary scale modeling (ESM) for the precise prediction of NPs. By integrating the GoogLeNet framework, NeuroScale effectively captures multi-scale NP features, enabling robust and accurate classification. Extensive benchmarking demonstrates its superior performance, consistently achieving an area under the receiver operating characteristic curve (AUC) exceeding 0.97. Additionally, we systematically analyzed the impact of protein sequence similarity thresholds and multi-scale sequence lengths on model performance, further validating NeuroScale's robustness and generalizability.

CONCLUSIONS

NeuroScale represents a significant advancement in neuropeptide prediction, offering both high accuracy and adaptability to diverse sequence characteristics. Its ability to generalize across different sequence similarity thresholds and lengths underscores its potential as a reliable tool for neuropeptide discovery and peptide-based drug development. By providing a scalable and efficient deep learning framework, NeuroScale paves the way for future research in neuropeptide function, disease mechanisms, and therapeutic applications.

Machine and deep learning to predict viral fusion peptides

Viral fusion proteins, located on the surface of enveloped viruses like SARS-CoV-2, Influenza, and HIV, play a vital role in fusing the virus envelope with the host cell membrane. Fusion peptides, conserved segments within these proteins, are crucial for the fusion process and are potential targets for therapy. Experimental identification of fusion peptides is time-consuming and costly, which creates the need for bioinformatics tools that can predict the segment within the fusion protein sequence that corresponds to the FP. Although homology-based methods have been used towards this end, they fail to identify fusion peptides lacking overall sequence similarity to known counterparts. Therefore, alternative methods are needed to discover new putative fusion peptides, namely those based on machine learning. In this study, we explore various ML-based approaches to identify fusion peptides within a fusion protein sequence. We employ token classification methods and sliding window approaches coupled with machine and deep learning models. We evaluate different protein sequence representations, including one-hot encoding, physicochemical features, as well as representations from Natural Language Processing, such as word embeddings and transformers. Through the examination of over 50 combinations of models and features, we achieve promising results, particularly with models based on a state-of-the-art transformer for amino acid token classification. Furthermore, we utilize the best models to predict hypothetical fusion peptides for SARS-CoV-2, and critically analyse annotated peptides from existing research. Overall, our models effectively predict the location of fusion peptides, even in viruses for which limited experimental data is available.

DeepNeuropePred: A robust and universal tool to predict cleavage sites from neuropeptide precursors by protein language model

Neuropeptides play critical roles in many biological processes such as growth, learning, memory, metabolism, and neuronal differentiation. A few approaches have been reported for predicting neuropeptides that are cleaved from precursor protein sequences. However, these models for cleavage site prediction of precursors were developed using a limited number of neuropeptide precursor datasets and simple precursors representation models. In addition, a universal method for predicting neuropeptide cleavage sites that can be applied to all species is still lacking. In this paper, we proposed a novel deep learning method called DeepNeuropePred, using a combination of pre-trained language model and Convolutional Neural Networks for feature extraction and predicting the neuropeptide cleavage sites from precursors. To demonstrate the model's effectiveness and robustness, we evaluated the performance of DeepNeuropePred and four models from the NeuroPred server in the independent dataset and our model achieved the highest AUC score (0.916), which are 6.9%, 7.8%, 8.8%, and 10.9% higher than Mammalian (0.857), insects (0.850), Mollusc (0.842) and Motif (0.826), respectively. For the convenience of researchers, we provide a web server (http://isyslab.info/NeuroPepV2/deepNeuropePred.jsp).

Phylogeny-aware linear B-cell epitope predictor detects targets associated with immune response to orthopoxviruses

We introduce a phylogeny-aware framework for predicting linear B-cell epitope (LBCE)-containing regions within proteins. Our approach leverages evolutionary information by using a taxonomic scaffold to build models trained on hierarchically structured data. The resulting models present performance equivalent or superior to generalist methods, despite using simpler features and a fraction of the data volume required by current state-of-the-art predictors. This allows the utilization of available data for major pathogen lineages to facilitate the prediction of LBCEs for emerging infectious agents. We demonstrate the efficacy of our approach by predicting new LBCEs in the monkeypox (MPXV) and vaccinia viruses. Experimental validation of selected targets using sera from infected patients confirms the presence of LBCEs, including candidates for the differential serodiagnosis of recent MPXV infections. These results point to the use of phylogeny-aware predictors as a useful strategy to facilitate the targeted development of immunodiagnostic tools.

DeepMineLys: Deep mining of phage lysins from human microbiome

Vast shotgun metagenomics data remain an underutilized resource for novel enzymes. Artificial intelligence (AI) has increasingly been applied to protein mining, but its conventional performance evaluation is interpolative in nature, and these trained models often struggle to extrapolate effectively when challenged with unknown data. In this study, we present a framework (DeepMineLys [deep mining of phage lysins from human microbiome]) based on the convolutional neural network (CNN) to identify phage lysins from three human microbiome datasets. When validated with an independent dataset, our method achieved an F1-score of 84.00%, surpassing existing methods by 20.84%. We expressed 16 lysin candidates from the top 100 sequences in E. coli, confirming 11 as active. The best one displayed an activity 6.2-fold that of lysozyme derived from hen egg white, establishing it as the most potent lysin from the human microbiome. Our study also underscores several important issues when applying AI to biology questions. This framework should be applicable for mining other proteins.

Privacy-Preserving Federated Learning With Domain Adaptation for Multi-Disease Ocular Disease Recognition

As one of the effective ways of ocular disease recognition, early fundus screening can help patients avoid unrecoverable blindness. Although deep learning is powerful for image-based ocular disease recognition, the performance mainly benefits from a large number of labeled data. For ocular disease, data collection and annotation in a single site usually take a lot of time. If multi-site data are obtained, there are two main issues: 1) the data privacy is easy to be leaked; 2) the domain gap among sites will influence the recognition performance. Inspired by the above, first, a Gaussian randomized mechanism is adopted in local sites, which are then engaged in a global model to preserve the data privacy of local sites and models. Second, to bridge the domain gap among different sites, a two-step domain adaptation method is introduced, which consists of a domain confusion module and a multi-expert learning strategy. Based on the above, a privacy-preserving federated learning framework with domain adaptation is constructed. In the experimental part, a multi-disease early fundus screening dataset, including a detailed ablation study and four experimental settings, is used to show the stepwise performance, which verifies the efficiency of our proposed framework.

Omics-based deep learning approaches for lung cancer decision-making and therapeutics development

Lung cancer has been the most common and the leading cause of cancer deaths globally. Besides clinicopathological observations and traditional molecular tests, the advent of robust and scalable techniques for nucleic acid analysis has revolutionized biological research and medicinal practice in lung cancer treatment. In response to the demands for minimally invasive procedures and technology development over the past decade, many types of multi-omics data at various genome levels have been generated. As omics data grow, artificial intelligence models, particularly deep learning, are prominent in developing more rapid and effective methods to potentially improve lung cancer patient diagnosis, prognosis and treatment strategy. This decade has seen genome-based deep learning models thriving in various lung cancer tasks, including cancer prediction, subtype classification, prognosis estimation, cancer molecular signatures identification, treatment response prediction and biomarker development. In this study, we summarized available data sources for deep-learning-based lung cancer mining and provided an update on recent deep learning models in lung cancer genomics. Subsequently, we reviewed the current issues and discussed future research directions of deep-learning-based lung cancer genomics research.

Protein Sorting Prediction

Many computational methods are available for predicting protein sorting in bacteria. When comparing them, it is important to know that they can be grouped into three fundamentally different approaches: signal-based, global property-based, and homology-based prediction. In this chapter, the strengths and drawbacks of each of these approaches are described through many examples of methods that predict secretion, integration into membranes, or subcellular locations in general. The aim of this chapter is to provide a user-level introduction to the field with a minimum of computational theory.

Transformer-based deep learning for predicting protein properties in the life sciences

Recent developments in deep learning, coupled with an increasing number of sequenced proteins, have led to a breakthrough in life science applications, in particular in protein property prediction. There is hope that deep learning can close the gap between the number of sequenced proteins and proteins with known properties based on lab experiments. Language models from the field of natural language processing have gained popularity for protein property predictions and have led to a new computational revolution in biology, where old prediction results are being improved regularly. Such models can learn useful multipurpose representations of proteins from large open repositories of protein sequences and can be used, for instance, to predict protein properties. The field of natural language processing is growing quickly because of developments in a class of models based on a particular model-the Transformer model. We review recent developments and the use of large-scale Transformer models in applications for predicting protein characteristics and how such models can be used to predict, for example, post-translational modifications. We review shortcomings of other deep learning models and explain how the Transformer models have quickly proven to be a very promising way to unravel information hidden in the sequences of amino acids.

Biomolecular Topology: Modelling and Analysis

With the great advancement of experimental tools, a tremendous amount of biomolecular data has been generated and accumulated in various databases. The high dimensionality, structural complexity, the nonlinearity, and entanglements of biomolecular data, ranging from DNA knots, RNA secondary structures, protein folding configurations, chromosomes, DNA origami, molecular assembly, to others at the macromolecular level, pose a severe challenge in their analysis and characterization. In the past few decades, mathematical concepts, models, algorithms, and tools from algebraic topology, combinatorial topology, computational topology, and topological data analysis, have demonstrated great power and begun to play an essential role in tackling the biomolecular data challenge. In this work, we introduce biomolecular topology, which concerns the topological problems and models originated from the biomolecular systems. More specifically, the biomolecular topology encompasses topological structures, properties and relations that are emerged from biomolecular structures, dynamics, interactions, and functions. We discuss the various types of biomolecular topology from structures (of proteins, DNAs, and RNAs), protein folding, and protein assembly. A brief discussion of databanks (and databases), theoretical models, and computational algorithms, is presented. Further, we systematically review related topological models, including graphs, simplicial complexes, persistent homology, persistent Laplacians, de Rham-Hodge theory, Yau-Hausdorff distance, and the topology-based machine learning models.

Hierarchical deep learning for predicting GO annotations by integrating protein knowledge

MOTIVATION

Experimental testing and manual curation are the most precise ways for assigning Gene Ontology (GO) terms describing protein functions. However, they are expensive, time-consuming and cannot cope with the exponential growth of data generated by high-throughput sequencing methods. Hence, researchers need reliable computational systems to help fill the gap with automatic function prediction. The results of the last Critical Assessment of Function Annotation challenge revealed that GO-terms prediction remains a very challenging task. Recent developments on deep learning are significantly breaking out the frontiers leading to new knowledge in protein research thanks to the integration of data from multiple sources. However, deep models hitherto developed for functional prediction are mainly focused on sequence data and have not achieved breakthrough performances yet.

RESULTS

We propose DeeProtGO, a novel deep-learning model for predicting GO annotations by integrating protein knowledge. DeeProtGO was trained for solving 18 different prediction problems, defined by the three GO sub-ontologies, the type of proteins, and the taxonomic kingdom. Our experiments reported higher prediction quality when more protein knowledge is integrated. We also benchmarked DeeProtGO against state-of-the-art methods on public datasets, and showed it can effectively improve the prediction of GO annotations.

AVAILABILITY AND IMPLEMENTATION

DeeProtGO and a case of use are available at https://github.com/gamerino/DeeProtGO.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Struct2Graph: a graph attention network for structure based predictions of protein-protein interactions

BACKGROUND

Development of new methods for analysis of protein-protein interactions (PPIs) at molecular and nanometer scales gives insights into intracellular signaling pathways and will improve understanding of protein functions, as well as other nanoscale structures of biological and abiological origins. Recent advances in computational tools, particularly the ones involving modern deep learning algorithms, have been shown to complement experimental approaches for describing and rationalizing PPIs. However, most of the existing works on PPI predictions use protein-sequence information, and thus have difficulties in accounting for the three-dimensional organization of the protein chains.

RESULTS

In this study, we address this problem and describe a PPI analysis based on a graph attention network, named Struct2Graph, for identifying PPIs directly from the structural data of folded protein globules. Our method is capable of predicting the PPI with an accuracy of 98.89% on the balanced set consisting of an equal number of positive and negative pairs. On the unbalanced set with the ratio of 1:10 between positive and negative pairs, Struct2Graph achieves a fivefold cross validation average accuracy of 99.42%. Moreover, Struct2Graph can potentially identify residues that likely contribute to the formation of the protein-protein complex. The identification of important residues is tested for two different interaction types: (a) Proteins with multiple ligands competing for the same binding area, (b) Dynamic protein-protein adhesion interaction. Struct2Graph identifies interacting residues with 30% sensitivity, 89% specificity, and 87% accuracy.

CONCLUSIONS

In this manuscript, we address the problem of prediction of PPIs using a first of its kind, 3D-structure-based graph attention network (code available at https://github.com/baranwa2/Struct2Graph ). Furthermore, the novel mutual attention mechanism provides insights into likely interaction sites through its unsupervised knowledge selection process. This study demonstrates that a relatively low-dimensional feature embedding learned from graph structures of individual proteins outperforms other modern machine learning classifiers based on global protein features. In addition, through the analysis of single amino acid variations, the attention mechanism shows preference for disease-causing residue variations over benign polymorphisms, demonstrating that it is not limited to interface residues.

Overview of methods for characterization and visualization of a protein-protein interaction network in a multi-omics integration context

At the heart of the cellular machinery through the regulation of cellular functions, protein-protein interactions (PPIs) have a significant role. PPIs can be analyzed with network approaches. Construction of a PPI network requires prediction of the interactions. All PPIs form a network. Different biases such as lack of data, recurrence of information, and false interactions make the network unstable. Integrated strategies allow solving these different challenges. These approaches have shown encouraging results for the understanding of molecular mechanisms, drug action mechanisms, and identification of target genes. In order to give more importance to an interaction, it is evaluated by different confidence scores. These scores allow the filtration of the network and thus facilitate the representation of the network, essential steps to the identification and understanding of molecular mechanisms. In this review, we will discuss the main computational methods for predicting PPI, including ones confirming an interaction as well as the integration of PPIs into a network, and we will discuss visualization of these complex data.

Hom-Complex-Based Machine Learning (HCML) for the Prediction of Protein-Protein Binding Affinity Changes upon Mutation

Protein-protein interactions (PPIs) are involved in almost all biological processes in the cell. Understanding protein-protein interactions holds the key for the understanding of biological functions, diseases and the development of therapeutics. Recently, artificial intelligence (AI) models have demonstrated great power in PPIs. However, a key issue for all AI-based PPI models is efficient molecular representations and featurization. Here, we propose Hom-complex-based PPI representation, and Hom-complex-based machine learning models for the prediction of PPI binding affinity changes upon mutation, for the first time. In our model, various Hom complexes Hom(G₁, G) can be generated for the graph representation G of protein-protein complex by using different graphs G₁, which reveal G₁-related inner connections within the graph representation G of protein-protein complex. Further, for a specific graph G₁, a series of nested Hom complexes are generated to give a multiscale characterization of the PPIs. Its persistent homology and persistent Euler characteristic are used as molecular descriptors and further combined with the machine learning model, in particular, gradient boosting tree (GBT). We systematically test our model on the two most-commonly used data sets, that is, SKEMPI and AB-Bind. It has been found that our model outperforms all the existing models as far as we know, which demonstrates the great potential of our model for the analysis of PPIs. Our model can be used for the analysis and design of efficient antibodies for SARS-CoV-2.

Predicting RNA solvent accessibility from multi-scale context feature via multi-shot neural network

Knowledge of RNA solvent accessibility has recently become attractive due to the increasing awareness of its importance for key biological process. Accurately predicting the solvent accessibility of RNA is crucial for understanding its 3D structure and biological function. In this study, we develop a novel computational method, termed M²pred, for accurately predicting the solvent accessibility of RNA from sequence-based multi-scale context feature. In M²pred, three single-view features, i.e., base-pairing probabilities, position-specific frequency matrix, and a binary one-hot encoding, are first generated as three feature sources, and immediately concatenated to engender a super feature. Secondly, for the super feature, the matrix-format features of each nucleotide are extracted using an initialized sliding window technique, and regularly stacked into a cube-format feature. Then, using multi-scale context feature extraction strategy, a pyramid feature constructed of contextual feature of four scales related to target nucleotides is extracted from the cube-format feature. Finally, a customized multi-shot neural network framework, which is equipped with four different scales of receptive fields mainly integrating several residual attention blocks, is designed to dig discrimination information from the contextual pyramid feature. Experimental results demonstrate that the proposed M²pred achieve a high prediction performance and outperforms existing state-of-the-art prediction methods of RNA solvent accessibility.

DeepIII: Predicting Isoform-Isoform Interactions by Deep Neural Networks and Data Fusion

Alternative splicing enables a gene translating into different isoforms and into the corresponding proteoforms, which actually accomplish various biological functions of a living body. Isoform-isoform interactions (IIIs) provide a higher resolution interactome to explore the cellular processes and disease mechanisms than the canonically studied protein-protein interactions (PPIs), which are often recorded at the coarse gene level. The knowledge of IIIs is critical to map pathways, understand protein complexity and functional diversity, but the known IIIs are very scanty. In this paper, we propose a deep learning based method called DeepIII to systematically predict genome-wide IIIs by integrating diverse data sources, including RNA-seq datasets of different human tissues, exon array data, domain-domain interactions (DDIs) of proteins, nucleotide sequences and amino acid sequences. Particularly, DeepIII fuses these data to learn the representation of isoform pairs with a four-layer deep neural networks, and then performs binary classification on the learnt representation to achieve the prediction of IIIs. Experimental results show that DeepIII achieves a superior prediction performance to the state-of-the-art solutions and the III network constructed by DeepIII gives more accurate isoform function prediction. Case studies further confirm that DeepIII can differentiate the individual interaction partners of different isoforms spliced from the same gene. The code and datasets of DeepIII are available at http://mlda.swu.edu.cn/codes.php?name=DeepIII.

Analysis of Data Interaction Process Based on Data Mining and Neural Network Topology Visualization

This paper addresses data mining and neural network model construction and analysis to design a data interaction process model based on data mining and topology visualization. This paper performs preprocessing data operations such as data filtering and cleaning of the collected data. A typical multichannel convolutional neural network (MCNN) in deep learning techniques is applied to alert students' academic performance. In addition, the network topology of the CNN is optimized to improve the performance of the model. The CNN has many hyperparameters that need to be tuned to construct an optimal model that can effectively interact with the data. In this paper, we propose a method to visualize the network topology within unstable regions to address the current problem of lacking an effective way to layout the network topology into specified areas. The technique transforms the network topology layout problem within the unstable region into a circular topology diffusion problem within a convex polygon, ensuring a clear, logical topology connection, and dramatically reducing the gaps in the area, making the layout more uniform beautiful. This paper constructs a real-time data interaction model based on JSON format and database triggers using message queues for reliable delivery. A platform-based real-time data interaction solution is designed by combining the timer method with the original key. The solution designed in this paper considers the real-time accuracy, security and reliability of data interaction. It satisfies the platform's initial and newly discovered requirements for data interaction.

Matrikines as mediators of tissue remodelling

Extracellular matrix (ECM) proteins confer biomechanical properties, maintain cell phenotype and mediate tissue repair (via release of sequestered cytokines and proteases). In contrast to intracellular proteomes, where proteins are monitored and replaced over short time periods, many ECM proteins function for years (decades in humans) without replacement. The longevity of abundant ECM proteins, such as collagen I and elastin, leaves them vulnerable to damage accumulation and their host organs prone to chronic, age-related diseases. However, ECM protein fragmentation can potentially produce peptide cytokines (matrikines) which may exacerbate and/or ameliorate age- and disease-related ECM remodelling. In this review, we discuss ECM composition, function and degradation and highlight examples of endogenous matrikines. We then critically and comprehensively analyse published studies of matrix-derived peptides used as topical skin treatments, before considering the potential for improvements in the discovery and delivery of novel matrix-derived peptides to skin and internal organs. From this, we conclude that while the translational impact of matrix-derived peptide therapeutics is evident, the mechanisms of action of these peptides are poorly defined. Further, well-designed, multimodal studies are required.

Differences in ligand-induced protein dynamics extracted from an unsupervised deep learning approach correlate with protein-ligand binding affinities

Prediction of protein–ligand binding affinity is a major goal in drug discovery. Generally, free energy gap is calculated between two states (e.g., ligand binding and unbinding). The energy gap implicitly includes the effects of changes in protein dynamics induced by ligand binding. However, the relationship between protein dynamics and binding affinity remains unclear. Here, we propose a method that represents ligand-binding-induced protein behavioral change with a simple feature that can be used to predict protein–ligand affinity. From unbiased molecular simulation data, an unsupervised deep learning method measures the differences in protein dynamics at a ligand-binding site depending on the bound ligands. A dimension reduction method extracts a dynamic feature that strongly correlates to the binding affinities. Moreover, the residues that play important roles in protein–ligand interactions are specified based on their contribution to the differences. These results indicate the potential for binding dynamics-based drug discovery.

Differences in ligand-induced protein dynamics extracted as a single feature from a deep learning-based analysis of MD simulations correlate with ligand binding affinity.

Improving the topology prediction of α-helical transmembrane proteins with deep transfer learning

Transmembrane proteins (TMPs) are essential for cell recognition and communication, and they serve as important drug targets in humans. Transmembrane proteins' 3D structures are critical for determining their functions and drug design but are hard to determine even by experimental methods. Although some computational methods have been developed to predict transmembrane helices (TMHs) and orientation, there is still room for improvement. Considering that the pre-trained language model can make full use of massive unlabeled protein sequences to obtain latent feature representation for TMPs and reduce the dependence on evolutionary information, we proposed DeepTMpred, which used pre-trained self-supervised language models called ESM, convolutional neural networks, attentive neural network and conditional random fields for alpha-TMP topology prediction. Compared with the current state-of-the-art tools on a non-redundant dataset of TMPs, DeepTMpred demonstrated superior predictive performance in most evaluation metrics, especially at the TMH level. Furthermore, DeepTMpred could also obtain reliable prediction results for TMPs without much evolutionary feature in a few seconds. A tutorial on how to use DeepTMpred can be found in the colab notebook (https://colab.research.google.com/github/ISYSLAB-HUST/DeepTMpred/blob/master/notebook/test.ipynb).

Psychosocial Factors and Psychological Characteristics of Personality of Patients with Chronic Diseases Using Artificial Intelligence Data Mining Technology and Wireless Network Cloud Service Platform

It was to explore the application value of health cloud service platform based on data mining algorithm and wireless network in the analysis of psychosocial factors and psychological characteristics of personality of patients with chronic diseases. Based on the demand analysis of cloud service platform for chronic diseases, a health cloud service platform including three modules was established: support layer, application layer, and interaction layer; and K-means algorithm and Apriori algorithm were used to mine and process data. The changes of pulse wave and EEG signal of epileptic seizures before and after processing by wireless network health cloud service platform were analyzed. 42 patients with idiopathic generalized epilepsy were selected as the research subjects, and 40 volunteers with normal physical examination during the same period were selected as the control group. The differences in the basic clinical characteristics data, Hamilton Anxiety Scale (HAMA), Hamilton Depression Scale (HAMD), Symptom Checklist 90 (SCL-90), and Eysenck Personality Questionnaire-Revision Short Scale for Chinese (EPQ-RSC) were compared between the two groups. It was found that the initial EEG signals of epileptic patients had noise pollution before and after the seizure, and the noise in the EEG signals was filtered out after digital technology processing in the cloud service platform. The maximum number of epileptic patients aged 18∼30 years was 17 (40.48%), and the mean scores of HAMD and HAMA scales in the epileptic group were significantly higher than those in the control group (P < 0.001). The total score of SCL-90, somatization, obsessive-compulsive symptoms, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation, and psychosis in the epilepsy group were obviously higher than those in the control group (P < 0.01). The mean value of EPQ-RSC and neuroticism (N) was clearly higher (P < 0.05), the mean value of extroversion (E) was significantly lower (P < 0.01), and the mean value of Lie Scale was significantly higher (P < 0.05) in the epileptic group in contrast with those in the control group. It indicates that the cloud service platform for chronic diseases based on artificial intelligence data mining technology and wireless network has potential application value. Epilepsy patients with chronic diseases should be paid more attention to their psychosocial factors and psychological characteristics of personality in the treatment process.

PFmulDL: a novel strategy enabling multi-class and multi-label protein function annotation by integrating diverse deep learning methods

Bioinformatic annotation of protein function is essential but extremely sophisticated, which asks for extensive efforts to develop effective prediction method. However, the existing methods tend to amplify the representativeness of the families with large number of proteins by misclassifying the proteins in the families with small number of proteins. That is to say, the ability of the existing methods to annotate proteins in the 'rare classes' remains limited. Herein, a new protein function annotation strategy, PFmulDL, integrating multiple deep learning methods, was thus constructed. First, the recurrent neural network was integrated, for the first time, with the convolutional neural network to facilitate the function annotation. Second, a transfer learning method was introduced to the model construction for further improving the prediction performances. Third, based on the latest data of Gene Ontology, the newly constructed model could annotate the largest number of protein families comparing with the existing methods. Finally, this newly constructed model was found capable of significantly elevating the prediction performance for the 'rare classes' without sacrificing that for the 'major classes'. All in all, due to the emerging requirements on improving the prediction performance for the proteins in 'rare classes', this new strategy would become an essential complement to the existing methods for protein function prediction. All the models and source codes are freely available and open to all users at: https://github.com/idrblab/PFmulDL.

VHH Structural Modelling Approaches: A Critical Review

VHH, i.e., VH domains of camelid single-chain antibodies, are very promising therapeutic agents due to their significant physicochemical advantages compared to classical mammalian antibodies. The number of experimentally solved VHH structures has significantly improved recently, which is of great help, because it offers the ability to directly work on 3D structures to humanise or improve them. Unfortunately, most VHHs do not have 3D structures. Thus, it is essential to find alternative ways to get structural information. The methods of structure prediction from the primary amino acid sequence appear essential to bypass this limitation. This review presents the most extensive overview of structure prediction methods applied for the 3D modelling of a given VHH sequence (a total of 21). Besides the historical overview, it aims at showing how model software programs have been shaping the structural predictions of VHHs. A brief explanation of each methodology is supplied, and pertinent examples of their usage are provided. Finally, we present a structure prediction case study of a recently solved VHH structure. According to some recent studies and the present analysis, AlphaFold 2 and NanoNet appear to be the best tools to predict a structural model of VHH from its sequence.

Persistent spectral based ensemble learning (PerSpect-EL) for protein-protein binding affinity prediction

Protein-protein interactions (PPIs) play a significant role in nearly all cellular and biological activities. Data-driven machine learning models have demonstrated great power in PPIs. However, the design of efficient molecular featurization poses a great challenge for all learning models for PPIs. Here, we propose persistent spectral (PerSpect) based PPI representation and featurization, and PerSpect-based ensemble learning (PerSpect-EL) models for PPI binding affinity prediction, for the first time. In our model, a sequence of Hodge (or combinatorial) Laplacian (HL) matrices at various different scales are generated from a specially designed filtration process. PerSpect attributes, which are statistical and combinatorial properties of spectrum information from these HL matrices, are used as features for PPI characterization. Each PerSpect attribute is input into a 1D convolutional neural network (CNN), and these CNN networks are stacked together in our PerSpect-based ensemble learning models. We systematically test our model on the two most commonly used datasets, i.e. SKEMPI and AB-Bind. It has been found that our model can achieve state-of-the-art results and outperform all existing models to the best of our knowledge.

PIPENN: protein interface prediction from sequence with an ensemble of neural nets

MOTIVATION

The interactions between proteins and other molecules are essential to many biological and cellular processes. Experimental identification of interface residues is a time-consuming, costly and challenging task, while protein sequence data are ubiquitous. Consequently, many computational and machine learning approaches have been developed over the years to predict such interface residues from sequence. However, the effectiveness of different Deep Learning (DL) architectures and learning strategies for protein-protein, protein-nucleotide and protein-small molecule interface prediction has not yet been investigated in great detail. Therefore, we here explore the prediction of protein interface residues using six DL architectures and various learning strategies with sequence-derived input features.

RESULTS

We constructed a large dataset dubbed BioDL, comprising protein-protein interactions from the PDB, and DNA/RNA and small molecule interactions from the BioLip database. We also constructed six DL architectures, and evaluated them on the BioDL benchmarks. This shows that no single architecture performs best on all instances. An ensemble architecture, which combines all six architectures, does consistently achieve peak prediction accuracy. We confirmed these results on the published benchmark set by Zhang and Kurgan (ZK448), and on our own existing curated homo- and heteromeric protein interaction dataset. Our PIPENN sequence-based ensemble predictor outperforms current state-of-the-art sequence-based protein interface predictors on ZK448 on all interaction types, achieving an AUC-ROC of 0.718 for protein-protein, 0.823 for protein-nucleotide and 0.842 for protein-small molecule.

AVAILABILITY AND IMPLEMENTATION

Source code and datasets are available at https://github.com/ibivu/pipenn/.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Protein pK a Prediction with Machine Learning

Protein pK _a prediction is essential for the investigation of the pH-associated relationship between protein structure and function. In this work, we introduce a deep learning-based protein pK _a predictor DeepKa, which is trained and validated with the pK _a values derived from continuous constant-pH molecular dynamics (CpHMD) simulations of 279 soluble proteins. Here, the CpHMD implemented in the Amber molecular dynamics package has been employed (Huang Y.J. Chem. Inf. Model.2018, 58, 1372-1383). Notably, to avoid discontinuities at the boundary, grid charges are proposed to represent protein electrostatics. We show that the prediction accuracy by DeepKa is close to that by CpHMD benchmarking simulations, validating DeepKa as an efficient protein pK _a predictor. In addition, the training and validation sets created in this study can be applied to the development of machine learning-based protein pK _a predictors in the future. Finally, the grid charge representation is general and applicable to other topics, such as the protein-ligand binding affinity prediction.

Facilitating Machine Learning-Guided Protein Engineering with Smart Library Design and Massively Parallel Assays

Protein design plays an important role in recent medical advances from antibody therapy to vaccine design. Typically, exhaustive mutational screens or directed evolution experiments are used for the identification of the best design or for improvements to the wild-type variant. Even with a high-throughput screening on pooled libraries and Next-Generation Sequencing to boost the scale of read-outs, surveying all the variants with combinatorial mutations for their empirical fitness scores is still of magnitudes beyond the capacity of existing experimental settings. To tackle this challenge, in-silico approaches using machine learning to predict the fitness of novel variants based on a subset of empirical measurements are now employed. These machine learning models turn out to be useful in many cases, with the premise that the experimentally determined fitness scores and the amino-acid descriptors of the models are informative. The machine learning models can guide the search for the highest fitness variants, resolve complex epistatic relationships, and highlight bio-physical rules for protein folding. Using machine learning-guided approaches, researchers can build more focused libraries, thus relieving themselves from labor-intensive screens and fast-tracking the optimization process. Here, we describe the current advances in massive-scale variant screens, and how machine learning and mutagenesis strategies can be integrated to accelerate protein engineering. More specifically, we examine strategies to make screens more economical, informative, and effective in discovery of useful variants.

Mapping the glycosyltransferase fold landscape using interpretable deep learning

Glycosyltransferases (GTs) play fundamental roles in nearly all cellular processes through the biosynthesis of complex carbohydrates and glycosylation of diverse protein and small molecule substrates. The extensive structural and functional diversification of GTs presents a major challenge in mapping the relationships connecting sequence, structure, fold and function using traditional bioinformatics approaches. Here, we present a convolutional neural network with attention (CNN-attention) based deep learning model that leverages simple secondary structure representations generated from primary sequences to provide GT fold prediction with high accuracy. The model learns distinguishing secondary structure features free of primary sequence alignment constraints and is highly interpretable. It delineates sequence and structural features characteristic of individual fold types, while classifying them into distinct clusters that group evolutionarily divergent families based on shared secondary structural features. We further extend our model to classify GT families of unknown folds and variants of known folds. By identifying families that are likely to adopt novel folds such as GT91, GT96 and GT97, our studies expand the GT fold landscape and prioritize targets for future structural studies.

DeepGOWeb: fast and accurate protein function prediction on the (Semantic) Web

Understanding the functions of proteins is crucial to understand biological processes on a molecular level. Many more protein sequences are available than can be investigated experimentally. DeepGOPlus is a protein function prediction method based on deep learning and sequence similarity. DeepGOWeb makes the prediction model available through a website, an API, and through the SPARQL query language for interoperability with databases that rely on Semantic Web technologies. DeepGOWeb provides accurate and fast predictions and ensures that predicted functions are consistent with the Gene Ontology; it can provide predictions for any protein and any function in Gene Ontology. DeepGOWeb is freely available at https://deepgo.cbrc.kaust.edu.sa/.

MolAICal: a soft tool for 3D drug design of protein targets by artificial intelligence and classical algorithm

Deep learning is an important branch of artificial intelligence that has been successfully applied into medicine and two-dimensional ligand design. The three-dimensional (3D) ligand generation in the 3D pocket of protein target is an interesting and challenging issue for drug design by deep learning. Here, the MolAICal software is introduced to supply a way for generating 3D drugs in the 3D pocket of protein targets by combining with merits of deep learning model and classical algorithm. The MolAICal software mainly contains two modules for 3D drug design. In the first module of MolAICal, it employs the genetic algorithm, deep learning model trained by FDA-approved drug fragments and Vinardo score fitting on the basis of PDBbind database for drug design. In the second module, it uses deep learning generative model trained by drug-like molecules of ZINC database and molecular docking invoked by Autodock Vina automatically. Besides, the Lipinski's rule of five, Pan-assay interference compounds (PAINS), synthetic accessibility (SA) and other user-defined rules are introduced for filtering out unwanted ligands in MolAICal. To show the drug design modules of MolAICal, the membrane protein glucagon receptor and non-membrane protein SARS-CoV-2 main protease are chosen as the investigative drug targets. The results show MolAICal can generate the various and novel ligands with good binding scores and appropriate XLOGP values. We believe that MolAICal can use the advantages of deep learning model and classical programming for designing 3D drugs in protein pocket. MolAICal is freely for any nonprofit purpose and accessible at https://molaical.github.io.

XGraphBoost: Extracting Graph Neural Network-Based Features for a Better Prediction of Molecular Properties

Determining the properties of chemical molecules is essential for screening candidates similar to a specific drug. These candidate molecules are further evaluated for their target binding affinities, side effects, target missing probabilities, etc. Conventional machine learning algorithms demonstrated satisfying prediction accuracies of molecular properties. A molecule cannot be directly loaded into a machine learning model, and a set of engineered features needs to be designed and calculated from a molecule. Such hand-crafted features rely heavily on the experiences of the investigating researchers. The concept of graph neural networks (GNNs) was recently introduced to describe the chemical molecules. The features may be automatically and objectively extracted from the molecules through various types of GNNs, e.g., GCN (graph convolution network), GGNN (gated graph neural network), DMPNN (directed message passing neural network), etc. However, the training of a stable GNN model requires a huge number of training samples and a large amount of computing power, compared with the conventional machine learning strategies. This study proposed the integrated framework XGraphBoost to extract the features using a GNN and build an accurate prediction model of molecular properties using the classifier XGBoost. The proposed framework XGraphBoost fully inherits the merits of the GNN-based automatic molecular feature extraction and XGBoost-based accurate prediction performance. Both classification and regression problems were evaluated using the framework XGraphBoost. The experimental results strongly suggest that XGraphBoost may facilitate the efficient and accurate predictions of various molecular properties. The source code is freely available to academic users at https://github.com/chenxiaowei-vincent/XGraphBoost.git.

nhKcr: a new bioinformatics tool for predicting crotonylation sites on human nonhistone proteins based on deep learning

Lysine crotonylation (Kcr) is a newly discovered type of protein post-translational modification and has been reported to be involved in various pathophysiological processes. High-resolution mass spectrometry is the primary approach for identification of Kcr sites. However, experimental approaches for identifying Kcr sites are often time-consuming and expensive when compared with computational approaches. To date, several predictors for Kcr site prediction have been developed, most of which are capable of predicting crotonylation sites on either histones alone or mixed histone and nonhistone proteins together. These methods exhibit high diversity in their algorithms, encoding schemes, feature selection techniques and performance assessment strategies. However, none of them were designed for predicting Kcr sites on nonhistone proteins. Therefore, it is desirable to develop an effective predictor for identifying Kcr sites from the large amount of nonhistone sequence data. For this purpose, we first provide a comprehensive review on six methods for predicting crotonylation sites. Second, we develop a novel deep learning-based computational framework termed as CNNrgb for Kcr site prediction on nonhistone proteins by integrating different types of features. We benchmark its performance against multiple commonly used machine learning classifiers (including random forest, logitboost, naïve Bayes and logistic regression) by performing both 10-fold cross-validation and independent test. The results show that the proposed CNNrgb framework achieves the best performance with high computational efficiency on large datasets. Moreover, to facilitate users' efforts to investigate Kcr sites on human nonhistone proteins, we implement an online server called nhKcr and compare it with other existing tools to illustrate the utility and robustness of our method. The nhKcr web server and all the datasets utilized in this study are freely accessible at http://nhKcr.erc.monash.edu/.

Predicting Proteolysis in Complex Proteomes Using Deep Learning

Both protease- and reactive oxygen species (ROS)-mediated proteolysis are thought to be key effectors of tissue remodeling. We have previously shown that comparison of amino acid composition can predict the differential susceptibilities of proteins to photo-oxidation. However, predicting protein susceptibility to endogenous proteases remains challenging. Here, we aim to develop bioinformatics tools to (i) predict cleavage site locations (and hence putative protein susceptibilities) and (ii) compare the predicted vulnerabilities of skin proteins to protease- and ROS-mediated proteolysis. The first goal of this study was to experimentally evaluate the ability of existing protease cleavage site prediction models (PROSPER and DeepCleave) to identify experimentally determined MMP9 cleavage sites in two purified proteins and in a complex human dermal fibroblast-derived extracellular matrix (ECM) proteome. We subsequently developed deep bidirectional recurrent neural network (BRNN) models to predict cleavage sites for 14 tissue proteases. The predictions of the new models were tested against experimental datasets and combined with amino acid composition analysis (to predict ultraviolet radiation (UVR)/ROS susceptibility) in a new web app: the Manchester proteome susceptibility calculator (MPSC). The BRNN models performed better in predicting cleavage sites in native dermal ECM proteins than existing models (DeepCleave and PROSPER), and application of MPSC to the skin proteome suggests that: compared with the elastic fiber network, fibrillar collagens may be susceptible primarily to protease-mediated proteolysis. We also identify additional putative targets of oxidative damage (dermatopontin, fibulins and defensins) and protease action (laminins and nidogen). MPSC has the potential to identify potential targets of proteolysis in disparate tissues and disease states.

Artificial Intelligence in Drug Discovery: A Comprehensive Review of Data-driven and Machine Learning Approaches

As expenditure on drug development increases exponentially, the overall drug discovery process requires a sustainable revolution. Since artificial intelligence (AI) is leading the fourth industrial revolution, AI can be considered as a viable solution for unstable drug research and development. Generally, AI is applied to fields with sufficient data such as computer vision and natural language processing, but there are many efforts to revolutionize the existing drug discovery process by applying AI. This review provides a comprehensive, organized summary of the recent research trends in AI-guided drug discovery process including target identification, hit identification, ADMET prediction, lead optimization, and drug repositioning. The main data sources in each field are also summarized in this review. In addition, an in-depth analysis of the remaining challenges and limitations will be provided, and proposals for promising future directions in each of the aforementioned areas.

Neural networks for protein structure and function prediction and dynamic analysis

Hardware and software advancements along with the accumulation of large amounts of data in recent years have together spurred a remarkable growth in the application of neural networks to various scientific fields. Machine learning based on neural networks with multiple (hidden) layers is becoming an extremely powerful approach for analyzing data. With the accumulation of large amounts of protein data such as structural and functional assay data, the effects of such approaches within the field of protein informatics are increasing. Here, we introduce our recent studies based on applications of neural networks for protein structure and function prediction and dynamic analysis involving: (i) inter-residue contact prediction based on a multiple sequence alignment (MSA) of amino acid sequences, (ii) prediction of protein-compound interaction using assay data, and (iii) detection of protein allostery from trajectories of molecular dynamic (MD) simulation.