DeepGO: predicting protein functions from sequence and interactions using a deep ontology-aware classifier

Structure-based protein and small molecule generation using EGNN and diffusion models: A comprehensive review

Recent breakthroughs in deep learning have revolutionized protein sequence and structure prediction. These advancements are built on decades of protein design efforts, and are overcoming traditional time and cost limitations. Diffusion models, at the forefront of these innovations, significantly enhance design efficiency by automating knowledge acquisition. In the field of de novo protein design, the goal is to create entirely novel proteins with predetermined structures. Given the arbitrary positions of proteins in 3-D space, graph representations and their properties are widely used in protein generation studies. A critical requirement in protein modelling is maintaining spatial relationships under transformations (rotations, translations, and reflections). This property, known as equivariance, ensures that predicted protein characteristics adapt seamlessly to changes in orientation or position. Equivariant graph neural networks offer a solution to this challenge. By incorporating equivariant graph neural networks to learn the score of the probability density function in diffusion models, one can generate proteins with robust 3-D structural representations. This review examines the latest deep learning advancements, specifically focusing on frameworks that combine diffusion models with equivariant graph neural networks for protein generation.

Function-Genes and Disease-Genes Prediction Based on Network Embedding and One-Class Classification

Using genes which have been experimentally-validated for diseases (functions) can develop machine learning methods to predict new disease/function-genes. However, the prediction of both function-genes and disease-genes faces the same problem: there are only certain positive examples, but no negative examples. To solve this problem, we proposed a function/disease-genes prediction algorithm based on network embedding (Variational Graph Auto-Encoders, VGAE) and one-class classification (Fast Minimum Covariance Determinant, Fast-MCD): VGAEMCD. Firstly, we constructed a protein-protein interaction (PPI) network centered on experimentally-validated genes; then VGAE was used to get the embeddings of nodes (genes) in the network; finally, the embeddings were input into the improved deep learning one-class classifier based on Fast-MCD to predict function/disease-genes. VGAEMCD can predict function-gene and disease-gene in a unified way, and only the experimentally-verified genes are needed to provide (no need for expression profile). VGAEMCD outperforms classical one-class classification algorithms in Recall, Precision, F-measure, Specificity, and Accuracy. Further experiments show that seven metrics of VGAEMCD are higher than those of state-of-art function/disease-genes prediction algorithms. The above results indicate that VGAEMCD can well learn the distribution characteristics of positive examples and accurately identify function/disease-genes.

Decoding functional proteome information in model organisms using protein language models

Protein language models have been tested and proved to be reliable when used on curated datasets but have not yet been applied to full proteomes. Accordingly, we tested how two different machine learning-based methods performed when decoding functional information from the proteomes of selected model organisms. We found that protein language models are more precise and informative than deep learning methods for all the species tested and across the three gene ontologies studied, and that they better recover functional information from transcriptomic experiments. The results obtained indicate that these language models are likely to be suitable for large-scale annotation and downstream analyses, and we recommend a guide for their use.

Precise metabolic modeling in post-omics era: accomplishments and perspectives

Microbes have been extensively utilized for their sustainable and scalable properties in synthesizing desired bio-products. However, insufficient knowledge about intracellular metabolism has impeded further microbial applications. The genome-scale metabolic models (GEMs) play a pivotal role in facilitating a global understanding of cellular metabolic mechanisms. These models enable rational modification by exploring metabolic pathways and predicting potential targets in microorganisms, enabling precise cell regulation without experimental costs. Nonetheless, simplified GEM only considers genome information and network stoichiometry while neglecting other important bio-information, such as enzyme functions, thermodynamic properties, and kinetic parameters. Consequently, uncertainties persist particularly when predicting microbial behaviors in complex and fluctuant systems. The advent of the omics era with its massive quantification of genes, proteins, and metabolites under various conditions has led to the flourishing of multi-constrained models and updated algorithms with improved predicting power and broadened dimension. Meanwhile, machine learning (ML) has demonstrated exceptional analytical and predictive capacities when applied to training sets of biological big data. Incorporating the discriminant strength of ML with GEM facilitates mechanistic modeling efficiency and improves predictive accuracy. This paper provides an overview of research innovations in the GEM, including multi-constrained modeling, analytical approaches, and the latest applications of ML, which may contribute comprehensive knowledge toward genetic refinement, strain development, and yield enhancement for a broad range of biomolecules.

Accurate prediction of protein function using statistics-informed graph networks

Understanding protein function is pivotal in comprehending the intricate mechanisms that underlie many crucial biological activities, with far-reaching implications in the fields of medicine, biotechnology, and drug development. However, more than 200 million proteins remain uncharacterized, and computational efforts heavily rely on protein structural information to predict annotations of varying quality. Here, we present a method that utilizes statistics-informed graph networks to predict protein functions solely from its sequence. Our method inherently characterizes evolutionary signatures, allowing for a quantitative assessment of the significance of residues that carry out specific functions. PhiGnet not only demonstrates superior performance compared to alternative approaches but also narrows the sequence-function gap, even in the absence of structural information. Our findings indicate that applying deep learning to evolutionary data can highlight functional sites at the residue level, providing valuable support for interpreting both existing properties and new functionalities of proteins in research and biomedicine.

DeepPRMS: advanced deep learning model to predict protein arginine methylation sites

Protein methylation is a form of post-translational modifications of protein, which is crucial for various cellular processes, including transcription activity and DNA repair. Correctly predicting protein methylation sites is fundamental for research and drug discovery. Some experimental techniques, such as methyl-specific antibodies, chromatin immune precipitation and mass spectrometry, exist for predicting protein methylation sites, but these techniques are time-consuming and costly. The ability to predict methylation sites using in silico techniques may help researchers identify potential candidate sites for future examination and make it easier to carry out site-specific investigations and downstream characterizations. In this research, we proposed a novel deep learning-based predictor, named DeepPRMS, to identify protein methylation sites in primary sequences. The DeepPRMS utilizes the gated recurrent unit (GRU) and convolutional neural network (CNN) algorithms to extract the sequential and spatial information from the primary sequences. GRU is used to extract sequential information, while CNN is used for spatial information. We combined the latent representation of GRU and CNN models to have a better interaction among them. Based on the independent test data set, DeepPRMS obtained an accuracy of 85.32%, a specificity of 84.94%, Matthew's correlation coefficient of 0.71 and a sensitivity of 85.80%. The results indicate that DeepPRMS can predict protein methylation sites with high accuracy and outperform the state-of-the-art models. The DeepPRMS is expected to effectively guide future research experiments for identifying potential methylated protein sites. The web server is available at http://deepprms.nitsri.ac.in/.

Deep learning methods for protein function prediction

Predicting protein function from protein sequence, structure, interaction, and other relevant information is important for generating hypotheses for biological experiments and studying biological systems, and therefore has been a major challenge in protein bioinformatics. Numerous computational methods had been developed to advance protein function prediction gradually in the last two decades. Particularly, in the recent years, leveraging the revolutionary advances in artificial intelligence (AI), more and more deep learning methods have been developed to improve protein function prediction at a faster pace. Here, we provide an in-depth review of the recent developments of deep learning methods for protein function prediction. We summarize the significant advances in the field, identify several remaining major challenges to be tackled, and suggest some potential directions to explore. The data sources and evaluation metrics widely used in protein function prediction are also discussed to assist the machine learning, AI, and bioinformatics communities to develop more cutting-edge methods to advance protein function prediction.

Prediction of DNA origami shape using graph neural network

Unlike proteins, which have a wealth of validated structural data, experimentally or computationally validated DNA origami datasets are limited. Here we present a graph neural network that can predict the three-dimensional conformation of DNA origami assemblies both rapidly and accurately. We develop a hybrid data-driven and physics-informed approach for model training, designed to minimize not only the data-driven loss but also the physics-informed loss. By employing an ensemble strategy, the model can successfully infer the shape of monomeric DNA origami structures almost in real time. Further refinement of the model in an unsupervised manner enables the analysis of supramolecular assemblies consisting of tens to hundreds of DNA blocks. The proposed model enables an automated inverse design of DNA origami structures for given target shapes. Our approach facilitates the real-time virtual prototyping of DNA origami, broadening its design space.

DualNetGO: a dual network model for protein function prediction via effective feature selection

MOTIVATION

Protein-protein interaction (PPI) networks are crucial for automatically annotating protein functions. As multiple PPI networks exist for the same set of proteins that capture properties from different aspects, it is a challenging task to effectively utilize these heterogeneous networks. Recently, several deep learning models have combined PPI networks from all evidence, or concatenated all graph embeddings for protein function prediction. However, the lack of a judicious selection procedure prevents the effective harness of information from different PPI networks, as these networks vary in densities, structures, and noise levels. Consequently, combining protein features indiscriminately could increase the noise level, leading to decreased model performance.

RESULTS

We develop DualNetGO, a dual-network model comprised of a Classifier and a Selector, to predict protein functions by effectively selecting features from different sources including graph embeddings of PPI networks, protein domain, and subcellular location information. Evaluation of DualNetGO on human and mouse datasets in comparison with other network-based models shows at least 4.5%, 6.2%, and 14.2% improvement on Fmax in BP, MF, and CC gene ontology categories, respectively, for human, and 3.3%, 10.6%, and 7.7% improvement on Fmax for mouse. We demonstrate the generalization capability of our model by training and testing on the CAFA3 data, and show its versatility by incorporating Esm2 embeddings. We further show that our model is insensitive to the choice of graph embedding method and is time- and memory-saving. These results demonstrate that combining a subset of features including PPI networks and protein attributes selected by our model is more effective in utilizing PPI network information than only using one kind of or concatenating graph embeddings from all kinds of PPI networks.

AVAILABILITY AND IMPLEMENTATION

The source code of DualNetGO and some of the experiment data are available at: https://github.com/georgedashen/DualNetGO.

Predicting protein functions using positive-unlabeled ranking with ontology-based priors

Automated protein function prediction is a crucial and widely studied problem in bioinformatics. Computationally, protein function is a multilabel classification problem where only positive samples are defined and there is a large number of unlabeled annotations. Most existing methods rely on the assumption that the unlabeled set of protein function annotations are negatives, inducing the false negative issue, where potential positive samples are trained as negatives. We introduce a novel approach named PU-GO, wherein we address function prediction as a positive-unlabeled ranking problem. We apply empirical risk minimization, i.e. we minimize the classification risk of a classifier where class priors are obtained from the Gene Ontology hierarchical structure. We show that our approach is more robust than other state-of-the-art methods on similarity-based and time-based benchmark datasets.

AVAILABILITY AND IMPLEMENTATION

Data and code are available at https://github.com/bio-ontology-research-group/PU-GO.

Machine learning integrative approaches to advance computational immunology

The study of immunology, traditionally reliant on proteomics to evaluate individual immune cells, has been revolutionized by single-cell RNA sequencing. Computational immunologists play a crucial role in analysing these datasets, moving beyond traditional protein marker identification to encompass a more detailed view of cellular phenotypes and their functional roles. Recent technological advancements allow the simultaneous measurements of multiple cellular components-transcriptome, proteome, chromatin, epigenetic modifications and metabolites-within single cells, including in spatial contexts within tissues. This has led to the generation of complex multiscale datasets that can include multimodal measurements from the same cells or a mix of paired and unpaired modalities. Modern machine learning (ML) techniques allow for the integration of multiple "omics" data without the need for extensive independent modelling of each modality. This review focuses on recent advancements in ML integrative approaches applied to immunological studies. We highlight the importance of these methods in creating a unified representation of multiscale data collections, particularly for single-cell and spatial profiling technologies. Finally, we discuss the challenges of these holistic approaches and how they will be instrumental in the development of a common coordinate framework for multiscale studies, thereby accelerating research and enabling discoveries in the computational immunology field.

POLAT: Protein function prediction based on soft mask graph network and residue-Label ATtention

MOTIVATION

Elucidating protein function is a central problem in biochemistry, genetics, and molecular biology. Developing computational methods for protein function prediction is critical due to the significant gap between sequence and functional data. Recent advances in protein structure prediction, which strongly correlates with function, make it feasible to use structure to predict function. However, current structure-based methods overlook the fact that individual residues may contribute differently to the protein's function and do not take into account the correlation between protein residues and their functions. The challenge of effectively utilizing the relationship between protein residues and function-level information to predict protein function remains unsolved.

RESULT

We proposed a protein function prediction method based on Soft Mask Graph Networks and Residue-Label Attention (POLAT), which could combine sequence features, predicted structure features, and function-level information to get an accurate prediction. We use soft mask graph networks to adaptively extract the residues relevant to functions. A residue-label attention mechanism is adopted to obtain the protein-level encoded features of a protein, which are then concatenated with a protein-level embedding and fed into a dense classifier to determine the probabilities of each function. POLAT achieves 0.670, 0.515, 0.578 Fmax and 0.677, 0.409, 0.507 AUPR on the PDB cdhit test set for the MFO, BPO, and CCO domains, respectively, outperforming the existing structure-based SOTA method GAT-GO (Fmax 0.633, 0.492, 0.547; AUPR 0.660, 0.381, 0.479). POLAT is also competitive in extensive experiments among sequence-based and multimodal methods and achieves the SOTA performance in three out of six metrics.

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.

Learning peptide properties with positive examples only

Deep learning can create accurate predictive models by exploiting existing large-scale experimental data, and guide the design of molecules. However, a major barrier is the requirement of both positive and negative examples in the classical supervised learning frameworks. Notably, most peptide databases come with missing information and low number of observations on negative examples, as such sequences are hard to obtain using high-throughput screening methods. To address this challenge, we solely exploit the limited known positive examples in a semi-supervised setting, and discover peptide sequences that are likely to map to certain antimicrobial properties via positive-unlabeled learning (PU). In particular, we use the two learning strategies of adapting base classifier and reliable negative identification to build deep learning models for inferring solubility, hemolysis, binding against SHP-2, and non-fouling activity of peptides, given their sequence. We evaluate the predictive performance of our PU learning method and show that by only using the positive data, it can achieve competitive performance when compared with the classical positive-negative (PN) classification approach, where there is access to both positive and negative examples.

AlphaFun: Structural-Alignment-Based Proteome Annotation Reveals why the Functionally Unknown Proteins (uPE1) Are So Understudied

With the rapid expansion of sequencing of genomes, the functional annotation of proteins becomes a bottleneck in understanding proteomes. The Chromosome-centric Human Proteome Project (C-HPP) aims to identify all proteins encoded by the human genome and find functional annotations for them. However, until now there are still 1137 identified human proteins without functional annotation, called uPE1 proteins. Sequence alignment was insufficient to predict their functions, and the crystal structures of most proteins were unavailable. In this study, we demonstrated a new functional annotation strategy, AlphaFun, based on structural alignment using deep-learning-predicted protein structures. Using this strategy, we functionally annotated 99% of the human proteome, including the uPE1 proteins and missing proteins, which have not been identified yet. The accuracy of the functional annotations was validated using the known-function proteins. The uPE1 proteins shared similar functions to the known-function PE1 proteins and tend to express only in very limited tissues. They are evolutionally young genes and thus should conduct functions only in specific tissues and conditions, limiting their occurrence in commonly studied biological models. Such functional annotations provide hints for functional investigations on the uPE1 proteins. This proteome-wide-scale functional annotation strategy is also applicable to any other species.

Protein function prediction through multi-view multi-label latent tensor reconstruction

BACKGROUND

In last two decades, the use of high-throughput sequencing technologies has accelerated the pace of discovery of proteins. However, due to the time and resource limitations of rigorous experimental functional characterization, the functions of a vast majority of them remain unknown. As a result, computational methods offering accurate, fast and large-scale assignment of functions to new and previously unannotated proteins are sought after. Leveraging the underlying associations between the multiplicity of features that describe proteins could reveal functional insights into the diverse roles of proteins and improve performance on the automatic function prediction task.

RESULTS

We present GO-LTR, a multi-view multi-label prediction model that relies on a high-order tensor approximation of model weights combined with non-linear activation functions. The model is capable of learning high-order relationships between multiple input views representing the proteins and predicting high-dimensional multi-label output consisting of protein functional categories. We demonstrate the competitiveness of our method on various performance measures. Experiments show that GO-LTR learns polynomial combinations between different protein features, resulting in improved performance. Additional investigations establish GO-LTR's practical potential in assigning functions to proteins under diverse challenging scenarios: very low sequence similarity to previously observed sequences, rarely observed and highly specific terms in the gene ontology.

IMPLEMENTATION

The code and data used for training GO-LTR is available at https://github.com/aalto-ics-kepaco/GO-LTR-prediction .

TinyNS: Platform-Aware Neurosymbolic Auto Tiny Machine Learning

Machine learning at the extreme edge has enabled a plethora of intelligent, time-critical, and remote applications. However, deploying interpretable artificial intelligence systems that can perform high-level symbolic reasoning and satisfy the underlying system rules and physics within the tight platform resource constraints is challenging. In this paper, we introduce TinyNS, the first platform-aware neurosymbolic architecture search framework for joint optimization of symbolic and neural operators. TinyNS provides recipes and parsers to automatically write microcontroller code for five types of neurosymbolic models, combining the context awareness and integrity of symbolic techniques with the robustness and performance of machine learning models. TinyNS uses a fast, gradient-free, black-box Bayesian optimizer over discontinuous, conditional, numeric, and categorical search spaces to find the best synergy of symbolic code and neural networks within the hardware resource budget. To guarantee deployability, TinyNS talks to the target hardware during the optimization process. We showcase the utility of TinyNS by deploying microcontroller-class neurosymbolic models through several case studies. In all use cases, TinyNS outperforms purely neural or purely symbolic approaches while guaranteeing execution on real hardware.

Effect of tokenization on transformers for biological sequences

MOTIVATION

Deep-learning models are transforming biological research, including many bioinformatics and comparative genomics algorithms, such as sequence alignments, phylogenetic tree inference, and automatic classification of protein functions. Among these deep-learning algorithms, models for processing natural languages, developed in the natural language processing (NLP) community, were recently applied to biological sequences. However, biological sequences are different from natural languages, such as English, and French, in which segmentation of the text to separate words is relatively straightforward. Moreover, biological sequences are characterized by extremely long sentences, which hamper their processing by current machine-learning models, notably the transformer architecture. In NLP, one of the first processing steps is to transform the raw text to a list of tokens. Deep-learning applications to biological sequence data mostly segment proteins and DNA to single characters. In this work, we study the effect of alternative tokenization algorithms on eight different tasks in biology, from predicting the function of proteins and their stability, through nucleotide sequence alignment, to classifying proteins to specific families.

RESULTS

We demonstrate that applying alternative tokenization algorithms can increase accuracy and at the same time, substantially reduce the input length compared to the trivial tokenizer in which each character is a token. Furthermore, applying these tokenization algorithms allows interpreting trained models, taking into account dependencies among positions. Finally, we trained these tokenizers on a large dataset of protein sequences containing more than 400 billion amino acids, which resulted in over a 3-fold decrease in the number of tokens. We then tested these tokenizers trained on large-scale data on the above specific tasks and showed that for some tasks it is highly beneficial to train database-specific tokenizers. Our study suggests that tokenizers are likely to be a critical component in future deep-network analysis of biological sequence data.

AVAILABILITY AND IMPLEMENTATION

Code, data, and trained tokenizers are available on https://github.com/technion-cs-nlp/BiologicalTokenizers.

From tradition to innovation: conventional and deep learning frameworks in genome annotation

Following the milestone success of the Human Genome Project, the 'Encyclopedia of DNA Elements (ENCODE)' initiative was launched in 2003 to unearth information about the numerous functional elements within the genome. This endeavor coincided with the emergence of numerous novel technologies, accompanied by the provision of vast amounts of whole-genome sequences, high-throughput data such as ChIP-Seq and RNA-Seq. Extracting biologically meaningful information from this massive dataset has become a critical aspect of many recent studies, particularly in annotating and predicting the functions of unknown genes. The core idea behind genome annotation is to identify genes and various functional elements within the genome sequence and infer their biological functions. Traditional wet-lab experimental methods still rely on extensive efforts for functional verification. However, early bioinformatics algorithms and software primarily employed shallow learning techniques; thus, the ability to characterize data and features learning was limited. With the widespread adoption of RNA-Seq technology, scientists from the biological community began to harness the potential of machine learning and deep learning approaches for gene structure prediction and functional annotation. In this context, we reviewed both conventional methods and contemporary deep learning frameworks, and highlighted novel perspectives on the challenges arising during annotation underscoring the dynamic nature of this evolving scientific landscape.

Insights into the inner workings of transformer models for protein function prediction

MOTIVATION

We explored how explainable artificial intelligence (XAI) can help to shed light into the inner workings of neural networks for protein function prediction, by extending the widely used XAI method of integrated gradients such that latent representations inside of transformer models, which were finetuned to Gene Ontology term and Enzyme Commission number prediction, can be inspected too.

RESULTS

The approach enabled us to identify amino acids in the sequences that the transformers pay particular attention to, and to show that these relevant sequence parts reflect expectations from biology and chemistry, both in the embedding layer and inside of the model, where we identified transformer heads with a statistically significant correspondence of attribution maps with ground truth sequence annotations (e.g. transmembrane regions, active sites) across many proteins.

AVAILABILITY AND IMPLEMENTATION

Source code can be accessed at https://github.com/markuswenzel/xai-proteins.

FEZF2 and AIRE1: An Evolutionary Trade-off in the Elimination of Auto-reactive T Cells in the Thymus

Autoimmune Regulator 1 (AIRE1) and Forebrain Embryonic Zinc Finger-Like Protein 2 (FEZF2) play pivotal roles in orchestrating the expression of tissue-restricted antigens (TRA) to facilitate the elimination of autoreactive T cells. AIRE1's presence in the gonads of various vertebrates has raised questions about its potential involvement in gene expression control for germline cell selection. Nevertheless, the evolutionary history of these genes has remained enigmatic, as has the rationale behind their apparent redundancy in vertebrates. Furthermore, the origin of the elimination process itself has remained elusive. To shed light on these mysteries, we conducted a comprehensive evolutionary analysis employing a range of tools, including multiple sequence alignment, phylogenetic tree construction, ancestral sequence reconstruction, and positive selection assessment. Our investigations revealed intriguing insights. AIRE1 homologs emerged during the divergence of T cells in higher vertebrates, signifying its role in this context. Conversely, FEZF2 exhibited multiple homologs spanning invertebrates, lampreys, and higher vertebrates. Ancestral sequence reconstruction demonstrated distinct origins for AIRE1 and FEZF2, underscoring that their roles in regulating TRA have evolved through disparate pathways. Furthermore, it became evident that both FEZF2 and AIRE1 govern a diverse repertoire of genes, encompassing ancient and more recently diverged targets. Notably, FEZF2 demonstrates expression in both vertebrate and invertebrate embryos and germlines, accentuating its widespread role. Intriguingly, FEZF2 harbors motifs associated with autophagy, such as DKFPHP, SYSELWKSSL, and SYSEL, a process integral to cell selection in invertebrates. Our findings suggest that FEZF2 initially emerged to regulate self-elimination in the gonads of invertebrates. As organisms evolved toward greater complexity, AIRE1 likely emerged to complement FEZF2's role, participating in the regulation of cell selection for elimination in both gonads and the thymus. This dynamic interplay between AIRE1 and FEZF2 underscores their multifaceted contributions to TRA expression regulation across diverse evolutionary contexts.

AnnoPRO: a strategy for protein function annotation based on multi-scale protein representation and a hybrid deep learning of dual-path encoding

Protein function annotation has been one of the longstanding issues in biological sciences, and various computational methods have been developed. However, the existing methods suffer from a serious long-tail problem, with a large number of GO families containing few annotated proteins. Herein, an innovative strategy named AnnoPRO was therefore constructed by enabling sequence-based multi-scale protein representation, dual-path protein encoding using pre-training, and function annotation by long short-term memory-based decoding. A variety of case studies based on different benchmarks were conducted, which confirmed the superior performance of AnnoPRO among available methods. Source code and models have been made freely available at: https://github.com/idrblab/AnnoPRO and https://zenodo.org/records/10012272.

Rapid proteome-wide prediction of lipid-interacting proteins through ligand-guided structural genomics

Lipids are primary metabolites that play essential roles in multiple cellular pathways. Alterations in lipid metabolism and transport are associated with infectious diseases and cancers. As such, proteins involved in lipid synthesis, trafficking, and modification, are targets for therapeutic intervention. The ability to rapidly detect these proteins can accelerate their biochemical and structural characterization. However, it remains challenging to identify lipid binding motifs in proteins due to a lack of conservation at the amino acids level. Therefore, new bioinformatic tools that can detect conserved features in lipid binding sites are necessary. Here, we present Structure-based Lipid-interacting Pocket Predictor (SLiPP), a structural bioinformatics algorithm that uses machine learning to detect protein cavities capable of binding to lipids in experimental and AlphaFold-predicted protein structures. SLiPP, which can be used at proteome-wide scales, predicts lipid binding pockets with an accuracy of 96.8% and a F1 score of 86.9%. Our analyses revealed that the algorithm relies on hydrophobicity-related features to distinguish lipid binding pockets from those that bind to other ligands. Use of the algorithm to detect lipid binding proteins in the proteomes of various bacteria, yeast, and human have produced hits annotated or verified as lipid binding proteins, and many other uncharacterized proteins whose functions are not discernable from sequence alone. Because of its ability to identify novel lipid binding proteins, SLiPP can spur the discovery of new lipid metabolic and trafficking pathways that can be targeted for therapeutic development.

A comprehensive computational benchmark for evaluating deep learning-based protein function prediction approaches

Proteins play an important role in life activities and are the basic units for performing functions. Accurately annotating functions to proteins is crucial for understanding the intricate mechanisms of life and developing effective treatments for complex diseases. Traditional biological experiments struggle to keep pace with the growing number of known proteins. With the development of high-throughput sequencing technology, a wide variety of biological data provides the possibility to accurately predict protein functions by computational methods. Consequently, many computational methods have been proposed. Due to the diversity of application scenarios, it is necessary to conduct a comprehensive evaluation of these computational methods to determine the suitability of each algorithm for specific cases. In this study, we present a comprehensive benchmark, BeProf, to process data and evaluate representative computational methods. We first collect the latest datasets and analyze the data characteristics. Then, we investigate and summarize 17 state-of-the-art computational methods. Finally, we propose a novel comprehensive evaluation metric, design eight application scenarios and evaluate the performance of existing methods on these scenarios. Based on the evaluation, we provide practical recommendations for different scenarios, enabling users to select the most suitable method for their specific needs. All of these servers can be obtained from https://csuligroup.com/BEPROF and https://github.com/CSUBioGroup/BEPROF.

ELISL: early-late integrated synthetic lethality prediction in cancer

MOTIVATION

Anti-cancer therapies based on synthetic lethality (SL) exploit tumour vulnerabilities for treatment with reduced side effects, by targeting a gene that is jointly essential with another whose function is lost. Computational prediction is key to expedite SL screening, yet existing methods are vulnerable to prevalent selection bias in SL data and reliant on cancer or tissue type-specific omics, which can be scarce. Notably, sequence similarity remains underexplored as a proxy for related gene function and joint essentiality.

RESULTS

We propose ELISL, Early-Late Integrated SL prediction with forest ensembles, using context-free protein sequence embeddings and context-specific omics from cell lines and tissue. Across eight cancer types, ELISL showed superior robustness to selection bias and recovery of known SL genes, as well as promising cross-cancer predictions. Co-occurring mutations in a BRCA gene and ELISL-predicted pairs from the HH, FGF, WNT, or NEIL gene families were associated with longer patient survival times, revealing therapeutic potential.

AVAILABILITY AND IMPLEMENTATION

Data: 10.6084/m9.figshare.23607558 & Code: github.com/joanagoncalveslab/ELISL.

A CNN-CBAM-BIGRU model for protein function prediction

Understanding a protein's function based solely on its amino acid sequence is a crucial but intricate task in bioinformatics. Traditionally, this challenge has proven difficult. However, recent years have witnessed the rise of deep learning as a powerful tool, achieving significant success in protein function prediction. Their strength lies in their ability to automatically learn informative features from protein sequences, which can then be used to predict the protein's function. This study builds upon these advancements by proposing a novel model: CNN-CBAM+BiGRU. It incorporates a Convolutional Block Attention Module (CBAM) alongside BiGRUs. CBAM acts as a spotlight, guiding the CNN to focus on the most informative parts of the protein data, leading to more accurate feature extraction. BiGRUs, a type of Recurrent Neural Network (RNN), excel at capturing long-range dependencies within the protein sequence, which are essential for accurate function prediction. The proposed model integrates the strengths of both CNN-CBAM and BiGRU. This study's findings, validated through experimentation, showcase the effectiveness of this combined approach. For the human dataset, the suggested method outperforms the CNN-BIGRU+ATT model by +1.0 % for cellular components, +1.1 % for molecular functions, and +0.5 % for biological processes. For the yeast dataset, the suggested method outperforms the CNN-BIGRU+ATT model by +2.4 % for the cellular component, +1.2 % for molecular functions, and +0.6 % for biological processes.

In silico protein function prediction: the rise of machine learning-based approaches

Proteins function as integral actors in essential life processes, rendering the realm of protein research a fundamental domain that possesses the potential to propel advancements in pharmaceuticals and disease investigation. Within the context of protein research, an imperious demand arises to uncover protein functionalities and untangle intricate mechanistic underpinnings. Due to the exorbitant costs and limited throughput inherent in experimental investigations, computational models offer a promising alternative to accelerate protein function annotation. In recent years, protein pre-training models have exhibited noteworthy advancement across multiple prediction tasks. This advancement highlights a notable prospect for effectively tackling the intricate downstream task associated with protein function prediction. In this review, we elucidate the historical evolution and research paradigms of computational methods for predicting protein function. Subsequently, we summarize the progress in protein and molecule representation as well as feature extraction techniques. Furthermore, we assess the performance of machine learning-based algorithms across various objectives in protein function prediction, thereby offering a comprehensive perspective on the progress within this field.

Enzyme function and evolution through the lens of bioinformatics

Enzymes have been shaped by evolution over billions of years to catalyse the chemical reactions that support life on earth. Dispersed in the literature, or organised in online databases, knowledge about enzymes can be structured in distinct dimensions, either related to their quality as biological macromolecules, such as their sequence and structure, or related to their chemical functions, such as the catalytic site, kinetics, mechanism, and overall reaction. The evolution of enzymes can only be understood when each of these dimensions is considered. In addition, many of the properties of enzymes only make sense in the light of evolution. We start this review by outlining the main paradigms of enzyme evolution, including gene duplication and divergence, convergent evolution, and evolution by recombination of domains. In the second part, we overview the current collective knowledge about enzymes, as organised by different types of data and collected in several databases. We also highlight some increasingly powerful computational tools that can be used to close gaps in understanding, in particular for types of data that require laborious experimental protocols. We believe that recent advances in protein structure prediction will be a powerful catalyst for the prediction of binding, mechanism, and ultimately, chemical reactions. A comprehensive mapping of enzyme function and evolution may be attainable in the near future.

Learning sequence, structure, and function representations of proteins with language models

The sequence-structure-function relationships that ultimately generate the diversity of extant observed proteins is complex, as proteins bridge the gap between multiple informational and physical scales involved in nearly all cellular processes. One limitation of existing protein annotation databases such as UniProt is that less than 1% of proteins have experimentally verified functions, and computational methods are needed to fill in the missing information. Here, we demonstrate that a multi-aspect framework based on protein language models can learn sequence-structure-function representations of amino acid sequences, and can provide the foundation for sensitive sequence-structure-function aware protein sequence search and annotation. Based on this model, we introduce a multi-aspect information retrieval system for proteins, Protein-Vec, covering sequence, structure, and function aspects, that enables computational protein annotation and function prediction at tree-of-life scales.

Struct2GO: protein function prediction based on graph pooling algorithm and AlphaFold2 structure information

MOTIVATION

In recent years, there has been a breakthrough in protein structure prediction, and the AlphaFold2 model of the DeepMind team has improved the accuracy of protein structure prediction to the atomic level. Currently, deep learning-based protein function prediction models usually extract features from protein sequences and combine them with protein-protein interaction networks to achieve good results. However, for newly sequenced proteins that are not in the protein-protein interaction network, such models cannot make effective predictions. To address this, this article proposes the Struct2GO model, which combines protein structure and sequence data to enhance the precision of protein function prediction and the generality of the model.

RESULTS

We obtain amino acid residue embeddings in protein structure through graph representation learning, utilize the graph pooling algorithm based on a self-attention mechanism to obtain the whole graph structure features, and fuse them with sequence features obtained from the protein language model. The results demonstrate that compared with the traditional protein sequence-based function prediction model, the Struct2GO model achieves better results.

AVAILABILITY AND IMPLEMENTATION

The data underlying this article are available at https://github.com/lyjps/Struct2GO.

HNetGO: protein function prediction via heterogeneous network transformer

Protein function annotation is one of the most important research topics for revealing the essence of life at molecular level in the post-genome era. Current research shows that integrating multisource data can effectively improve the performance of protein function prediction models. However, the heavy reliance on complex feature engineering and model integration methods limits the development of existing methods. Besides, models based on deep learning only use labeled data in a certain dataset to extract sequence features, thus ignoring a large amount of existing unlabeled sequence data. Here, we propose an end-to-end protein function annotation model named HNetGO, which innovatively uses heterogeneous network to integrate protein sequence similarity and protein-protein interaction network information and combines the pretraining model to extract the semantic features of the protein sequence. In addition, we design an attention-based graph neural network model, which can effectively extract node-level features from heterogeneous networks and predict protein function by measuring the similarity between protein nodes and gene ontology term nodes. Comparative experiments on the human dataset show that HNetGO achieves state-of-the-art performance on cellular component and molecular function branches.

HiFun: homology independent protein function prediction by a novel protein-language self-attention model

Protein function prediction based on amino acid sequence alone is an extremely challenging but important task, especially in metagenomics/metatranscriptomics field, in which novel proteins have been uncovered exponentially from new microorganisms. Many of them are extremely low homology to known proteins and cannot be annotated with homology-based or information integrative methods. To overcome this problem, we proposed a Homology Independent protein Function annotation method (HiFun) based on a unified deep-learning model by reassembling the sequence as protein language. The robustness of HiFun was evaluated using the benchmark datasets and metrics in the CAFA3 challenge. To navigate the utility of HiFun, we annotated 2 212 663 unknown proteins and discovered novel motifs in the UHGP-50 catalog. We proved that HiFun can extract latent function related structure features which empowers it ability to achieve function annotation for non-homology proteins. HiFun can substantially improve newly proteins annotation and expand our understanding of microorganisms' adaptation in various ecological niches. Moreover, we provided a free and accessible webservice at http://www.unimd.org/HiFun, requiring only protein sequences as input, offering researchers an efficient and practical platform for predicting protein functions.

Prediction of hot spots towards drug discovery by protein sequence embedding with 1D convolutional neural network

Protein hotspot residues are key sites that mediate protein-protein interactions. Accurate identification of these residues is essential for understanding the mechanism from protein to function and for designing drug targets. Current research has mostly focused on using machine learning methods to predict hot spots from known interface residues, which artificially extract the corresponding features of amino acid residues from sequence, structure, evolution, energy, and other information to train and test machine learning models. The process is cumbersome, time-consuming and laborious to some extent. This paper proposes a novel idea that develops a pre-trained protein sequence embedding model combined with a one-dimensional convolutional neural network, called Embed-1dCNN, to predict protein hotspot residues. In order to obtain large data samples, this work integrates and extracts data from the datasets of ASEdb, BID, SKEMPI and dbMPIKT to generate a new dataset, and adopts the SMOTE algorithm to expand positive samples to form the training set. The experimental results show that the method achieves an F1 score of 0.82 on the test set. Compared with other hot spot prediction methods, our model achieved better prediction performance.

SERF, a family of tiny highly conserved, highly charged proteins with enigmatic functions

Amyloid formation is a misfolding process that has been linked to age-related diseases, including Alzheimer's and Huntington's. Understanding how cellular factors affect this process in vivo is vital in realizing the dream of controlling this insidious process that robs so many people of their humanity. SERF (small EDRK-rich factor) was initially isolated as a factor that accelerated polyglutamine amyloid formation in a C. elegans model. SERF knockouts inhibit amyloid formation of a number of proteins that include huntingtin, α-synuclein and β-amyloid which are associated with Huntington's, Parkinson's and Alzheimer's disease, respectively, and purified SERF protein speeds their amyloid formation in vitro. SERF proteins are highly conserved, highly charged and conformationally dynamic proteins that form a fuzzy complex with amyloid precursors. They appear to act by specifically accelerating the primary step of amyloid nucleation. Brain-specific SERF knockout mice, though viable, appear to be more prone to deposition of amyloids, and show modified fibril morphology. Whole-body knockouts are perinatally lethal due to an apparently unrelated developmental issue. Recently, it was found that SERF binds RNA and is localized to nucleic acid-rich membraneless compartments. SERF-related sequences are commonly found fused to zinc finger sequences. These results point towards a nucleic acid-binding function. How this function relates to their ability to accelerate amyloid formation is currently obscure. In this review, we discuss the possible biological functions of SERF family proteins in the context of their structural fuzziness, modulation of amyloid pathway, nucleic acid binding and their fusion to folded proteins.

Prot2GO: Predicting GO Annotations From Protein Sequences and Interactions

Protein is the main material basis of living organisms and plays crucial role in life activities. Understanding the function of protein is of great significance for new drug discovery, disease treatment and vaccine development. In recent years, with the widespread application of deep learning in bioinformatics, researchers have proposed many deep learning models to predict protein functions. However, the existing deep learning methods usually only consider protein sequences, and thus cannot effectively integrate multi-source data to annotate protein functions. In this article, we propose the Prot2GO model, which can integrate protein sequence and PPI network data to predict protein functions. We utilize an improved biased random walk algorithm to extract the features of PPI network. For sequence data, we use a convolutional neural network to obtain the local features of the sequence and a recurrent neural network to capture the long-range associations between amino acid residues in protein sequence. Moreover, Prot2GO adopts the attention mechanism to identify protein motifs and structural domains. Experiments show that Prot2GO model achieves the state-of-the-art performance on multiple metrics.

DeepSG2PPI: A Protein-Protein Interaction Prediction Method Based on Deep Learning

Protein-protein interaction (PPI) plays an important role in almost all life activities. Many protein interaction sites have been confirmed by biological experiments, but these PPI site identification methods are time-consuming and expensive. In this study, a deep learning-based PPI prediction method, named DeepSG2PPI, is developed. First, the protein sequence information is retrieved and the local context information of each amino acid residue is calculated. A two-dimensional convolutional neural network (2D-CNN) model is employed to extract features from a two-channel coding structure, in which an attention mechanism is embedded to assign higher weights to key features. Second, the global statistical information of each amino acid residue and the relationship graph between the protein and GO (Gene Ontology) function annotation are built, and the graph embedding vector is constructed to represent the biological features of the protein. Finally, a 2D-CNN model and two 1D-CNN models are combined for PPI prediction. The comparison analysis with existing algorithms shows that the DeepSG2PPI method has better performance. It provides more accurate and effective PPI site prediction, which will be helpful in reducing the cost and failure rate of biological experiments.

From Reductionistic Approach to Systems Immunology Approach for the Understanding of Tumor Microenvironment

The tumor microenvironment (TME) is a complex and dynamic ecosystem that includes a variety of immune cells mutually interacting with tumor cells, structural/stromal cells, and each other. The immune cells in the TME can have dual functions as pro-tumorigenic and anti-tumorigenic. To understand such paradoxical functions, the reductionistic approach classifies the immune cells into pro- and anti-tumor cells and suggests the therapeutic blockade of the pro-tumor and induction of the anti-tumor immune cells. This strategy has proven to be partially effective in prolonging patients' survival only in a fraction of patients without offering a cancer cure. Recent advances in multi-omics allow taking systems immunology approach. This essay discusses how a systems immunology approach could revolutionize our understanding of the TME by suggesting that internetwork interactions of the immune cell types create distinct collective functions independent of the function of each cellular constituent. Such collective function can be understood by the discovery of the immunological patterns in the TME and may be modulated as a therapeutic means for immunotherapy of cancer.

Patterns of transcription factor binding and epigenome at promoters allow interpretable predictability of multiple functions of non-coding and coding genes

Understanding the biological roles of all genes only through experimental methods is challenging. A computational approach with reliable interpretability is needed to infer the function of genes, particularly for non-coding RNAs. We have analyzed genomic features that are present across both coding and non-coding genes like transcription factor (TF) and cofactor ChIP-seq (823), histone modifications ChIP-seq (n = 621), cap analysis gene expression (CAGE) tags (n = 255), and DNase hypersensitivity profiles (n = 255) to predict ontology-based functions of genes. Our approach for gene function prediction was reliable (>90% balanced accuracy) for 486 gene-sets. PubMed abstract mining and CRISPR screens supported the inferred association of genes with biological functions, for which our method had high accuracy. Further analysis revealed that TF-binding patterns at promoters have high predictive strength for multiple functions. TF-binding patterns at the promoter add an unexplored dimension of explainable regulatory aspects of genes and their functions. Therefore, we performed a comprehensive analysis for the functional-specificity of TF-binding patterns at promoters and used them for clustering functions to reveal many latent groups of gene-sets involved in common major cellular processes. We also showed how our approach could be used to infer the functions of non-coding genes using the CRISPR screens of coding genes, which were validated using a long non-coding RNA CRISPR screen. Thus our results demonstrated the generality of our approach by using gene-sets from CRISPR screens. Overall, our approach opens an avenue for predicting the involvement of non-coding genes in various functions.

Discovering functionally important sites in proteins

Proteins play important roles in biology, biotechnology and pharmacology, and missense variants are a common cause of disease. Discovering functionally important sites in proteins is a central but difficult problem because of the lack of large, systematic data sets. Sequence conservation can highlight residues that are functionally important but is often convoluted with a signal for preserving structural stability. We here present a machine learning method to predict functional sites by combining statistical models for protein sequences with biophysical models of stability. We train the model using multiplexed experimental data on variant effects and validate it broadly. We show how the model can be used to discover active sites, as well as regulatory and binding sites. We illustrate the utility of the model by prospective prediction and subsequent experimental validation on the functional consequences of missense variants in HPRT1 which may cause Lesch-Nyhan syndrome, and pinpoint the molecular mechanisms by which they cause disease.

Large-scale predicting protein functions through heterogeneous feature fusion

As the volume of protein sequence and structure data grows rapidly, the functions of the overwhelming majority of proteins cannot be experimentally determined. Automated annotation of protein function at a large scale is becoming increasingly important. Existing computational prediction methods are typically based on expanding the relatively small number of experimentally determined functions to large collections of proteins with various clues, including sequence homology, protein-protein interaction, gene co-expression, etc. Although there has been some progress in protein function prediction in recent years, the development of accurate and reliable solutions still has a long way to go. Here we exploit AlphaFold predicted three-dimensional structural information, together with other non-structural clues, to develop a large-scale approach termed PredGO to annotate Gene Ontology (GO) functions for proteins. We use a pre-trained language model, geometric vector perceptrons and attention mechanisms to extract heterogeneous features of proteins and fuse these features for function prediction. The computational results demonstrate that the proposed method outperforms other state-of-the-art approaches for predicting GO functions of proteins in terms of both coverage and accuracy. The improvement of coverage is because the number of structures predicted by AlphaFold is greatly increased, and on the other hand, PredGO can extensively use non-structural information for functional prediction. Moreover, we show that over 205 000 ($\sim $100%) entries in UniProt for human are annotated by PredGO, over 186 000 ($\sim $90%) of which are based on predicted structure. The webserver and database are available at http://predgo.denglab.org/.

BioSeq-Diabolo: Biological sequence similarity analysis using Diabolo

As the key for biological sequence structure and function prediction, disease diagnosis and treatment, biological sequence similarity analysis has attracted more and more attentions. However, the exiting computational methods failed to accurately analyse the biological sequence similarities because of the various data types (DNA, RNA, protein, disease, etc) and their low sequence similarities (remote homology). Therefore, new concepts and techniques are desired to solve this challenging problem. Biological sequences (DNA, RNA and protein sequences) can be considered as the sentences of "the book of life", and their similarities can be considered as the biological language semantics (BLS). In this study, we are seeking the semantics analysis techniques derived from the natural language processing (NLP) to comprehensively and accurately analyse the biological sequence similarities. 27 semantics analysis methods derived from NLP were introduced to analyse biological sequence similarities, bringing new concepts and techniques to biological sequence similarity analysis. Experimental results show that these semantics analysis methods are able to facilitate the development of protein remote homology detection, circRNA-disease associations identification and protein function annotation, achieving better performance than the other state-of-the-art predictors in the related fields. Based on these semantics analysis methods, a platform called BioSeq-Diabolo has been constructed, which is named after a popular traditional sport in China. The users only need to input the embeddings of the biological sequence data. BioSeq-Diabolo will intelligently identify the task, and then accurately analyse the biological sequence similarities based on biological language semantics. BioSeq-Diabolo will integrate different biological sequence similarities in a supervised manner by using Learning to Rank (LTR), and the performance of the constructed methods will be evaluated and analysed so as to recommend the best methods for the users. The web server and stand-alone package of BioSeq-Diabolo can be accessed at http://bliulab.net/BioSeq-Diabolo/server/.

TEMPROT: protein function annotation using transformers embeddings and homology search

BACKGROUND

Although the development of sequencing technologies has provided a large number of protein sequences, the analysis of functions that each one plays is still difficult due to the efforts of laboratorial methods, making necessary the usage of computational methods to decrease this gap. As the main source of information available about proteins is their sequences, approaches that can use this information, such as classification based on the patterns of the amino acids and the inference based on sequence similarity using alignment tools, are able to predict a large collection of proteins. The methods available in the literature that use this type of feature can achieve good results, however, they present restrictions of protein length as input to their models. In this work, we present a new method, called TEMPROT, based on the fine-tuning and extraction of embeddings from an available architecture pre-trained on protein sequences. We also describe TEMPROT+, an ensemble between TEMPROT and BLASTp, a local alignment tool that analyzes sequence similarity, which improves the results of our former approach.

RESULTS

The evaluation of our proposed classifiers with the literature approaches has been conducted on our dataset, which was derived from CAFA3 challenge database. Both TEMPROT and TEMPROT+ achieved competitive results on [Formula: see text], [Formula: see text], AuPRC and IAuPRC metrics on Biological Process (BP), Cellular Component (CC) and Molecular Function (MF) ontologies compared to state-of-the-art models, with the main results equal to 0.581, 0.692 and 0.662 of [Formula: see text] on BP, CC and MF, respectively.

CONCLUSIONS

The comparison with the literature showed that our model presented competitive results compared the state-of-the-art approaches considering the amino acid sequence pattern recognition and homology analysis. Our model also presented improvements related to the input size that the model can use to train compared to the literature methods.

Learning Peptide Properties with Positive Examples Only

Deep learning can create accurate predictive models by exploiting existing large-scale experimental data, and guide the design of molecules. However, a major barrier is the requirement of both positive and negative examples in the classical supervised learning frameworks. Notably, most peptide databases come with missing information and low number of observations on negative examples, as such sequences are hard to obtain using high-throughput screening methods. To address this challenge, we solely exploit the limited known positive examples in a semi-supervised setting, and discover peptide sequences that are likely to map to certain antimicrobial properties via positive-unlabeled learning (PU). In particular, we use the two learning strategies of adapting base classifier and reliable negative identification to build deep learning models for inferring solubility, hemolysis, binding against SHP-2, and non-fouling activity of peptides, given their sequence. We evaluate the predictive performance of our PU learning method and show that by only using the positive data, it can achieve competitive performance when compared with the classical positive-negative (PN) classification approach, where there is access to both positive and negative examples.

MMSMAPlus: a multi-view multi-scale multi-attention embedding model for protein function prediction

Protein is the most important component in organisms and plays an indispensable role in life activities. In recent years, a large number of intelligent methods have been proposed to predict protein function. These methods obtain different types of protein information, including sequence, structure and interaction network. Among them, protein sequences have gained significant attention where methods are investigated to extract the information from different views of features. However, how to fully exploit the views for effective protein sequence analysis remains a challenge. In this regard, we propose a multi-view, multi-scale and multi-attention deep neural model (MMSMA) for protein function prediction. First, MMSMA extracts multi-view features from protein sequences, including one-hot encoding features, evolutionary information features, deep semantic features and overlapping property features based on physiochemistry. Second, a specific multi-scale multi-attention deep network model (MSMA) is built for each view to realize the deep feature learning and preliminary classification. In MSMA, both multi-scale local patterns and long-range dependence from protein sequences can be captured. Third, a multi-view adaptive decision mechanism is developed to make a comprehensive decision based on the classification results of all the views. To further improve the prediction performance, an extended version of MMSMA, MMSMAPlus, is proposed to integrate homology-based protein prediction under the framework of multi-view deep neural model. Experimental results show that the MMSMAPlus has promising performance and is significantly superior to the state-of-the-art methods. The source code can be found at https://github.com/wzy-2020/MMSMAPlus.

Graph Representation Learning and Its Applications: A Survey

Graphs are data structures that effectively represent relational data in the real world. Graph representation learning is a significant task since it could facilitate various downstream tasks, such as node classification, link prediction, etc. Graph representation learning aims to map graph entities to low-dimensional vectors while preserving graph structure and entity relationships. Over the decades, many models have been proposed for graph representation learning. This paper aims to show a comprehensive picture of graph representation learning models, including traditional and state-of-the-art models on various graphs in different geometric spaces. First, we begin with five types of graph embedding models: graph kernels, matrix factorization models, shallow models, deep-learning models, and non-Euclidean models. In addition, we also discuss graph transformer models and Gaussian embedding models. Second, we present practical applications of graph embedding models, from constructing graphs for specific domains to applying models to solve tasks. Finally, we discuss challenges for existing models and future research directions in detail. As a result, this paper provides a structured overview of the diversity of graph embedding models.

OncoRTT: Predicting novel oncology-related therapeutic targets using BERT embeddings and omics features

Late-stage drug development failures are usually a consequence of ineffective targets. Thus, proper target identification is needed, which may be possible using computational approaches. The reason being, effective targets have disease-relevant biological functions, and omics data unveil the proteins involved in these functions. Also, properties that favor the existence of binding between drug and target are deducible from the protein’s amino acid sequence. In this work, we developed OncoRTT, a deep learning (DL)-based method for predicting novel therapeutic targets. OncoRTT is designed to reduce suboptimal target selection by identifying novel targets based on features of known effective targets using DL approaches. First, we created the “OncologyTT” datasets, which include genes/proteins associated with ten prevalent cancer types. Then, we generated three sets of features for all genes: omics features, the proteins’ amino-acid sequence BERT embeddings, and the integrated features to train and test the DL classifiers separately. The models achieved high prediction performances in terms of area under the curve (AUC), i.e., AUC greater than 0.88 for all cancer types, with a maximum of 0.95 for leukemia. Also, OncoRTT outperformed the state-of-the-art method using their data in five out of seven cancer types commonly assessed by both methods. Furthermore, OncoRTT predicts novel therapeutic targets using new test data related to the seven cancer types. We further corroborated these results with other validation evidence using the Open Targets Platform and a case study focused on the top-10 predicted therapeutic targets for lung cancer.

CFAGO: cross-fusion of network and attributes based on attention mechanism for protein function prediction

MOTIVATION

Protein function annotation is fundamental to understanding biological mechanisms. The abundant genome-scale protein-protein interaction (PPI) networks, together with other protein biological attributes, provide rich information for annotating protein functions. As PPI networks and biological attributes describe protein functions from different perspectives, it is highly challenging to cross-fuse them for protein function prediction. Recently, several methods combine the PPI networks and protein attributes via the graph neural networks (GNNs). However, GNNs may inherit or even magnify the bias caused by noisy edges in PPI networks. Besides, GNNs with stacking of many layers may cause the over-smoothing problem of node representations.

RESULTS

We develop a novel protein function prediction method, CFAGO, to integrate single-species PPI networks and protein biological attributes via a multi-head attention mechanism. CFAGO is first pre-trained with an encoder-decoder architecture to capture the universal protein representation of the two sources. It is then fine-tuned to learn more effective protein representations for protein function prediction. Benchmark experiments on human and mouse datasets show CFAGO outperforms state-of-the-art single-species network-based methods by at least 7.59%, 6.90%, 11.68% in terms of m-AUPR, M-AUPR, and Fmax, respectively, demonstrating cross-fusion by multi-head attention mechanism can greatly improve the protein function prediction. We further evaluate the quality of captured protein representations in terms of Davies Bouldin Score, whose results show that cross-fused protein representations by multi-head attention mechanism are at least 2.7% better than that of original and concatenated representations. We believe CFAGO is an effective tool for protein function prediction.

AVAILABILITY AND IMPLEMENTATION

The source code of CFAGO and experiments data are available at: http://bliulab.net/CFAGO/.

A Deep Learning Framework for Predicting Protein Functions With Co-Occurrence of GO Terms

The understanding of protein functions is critical to many biological problems such as the development of new drugs and new crops. To reduce the huge gap between the increase of protein sequences and annotations of protein functions, many methods have been proposed to deal with this problem. These methods use Gene Ontology (GO) to classify the functions of proteins and consider one GO term as a class label. However, they ignore the co-occurrence of GO terms that is helpful for protein function prediction. We propose a new deep learning model, named DeepPFP-CO, which uses Graph Convolutional Network (GCN) to explore and capture the co-occurrence of GO terms to improve the protein function prediction performance. In this way, we can further deduce the protein functions by fusing the predicted propensity of the center function and its co-occurrence functions. We use Fmax and AUPR to evaluate the performance of DeepPFP-CO and compare DeepPFP-CO with state-of-the-art methods such as DeepGOPlus and DeepGOA. The computational results show that DeepPFP-CO outperforms DeepGOPlus and other methods. Moreover, we further analyze our model at the protein level. The results have demonstrated that DeepPFP-CO improves the performance of protein function prediction. DeepPFP-CO is available at https://csuligroup.com/DeepPFP/.

Computational prediction of disordered binding regions

One of the key features of intrinsically disordered regions (IDRs) is their ability to interact with a broad range of partner molecules. Multiple types of interacting IDRs were identified including molecular recognition fragments (MoRFs), short linear sequence motifs (SLiMs), and protein-, nucleic acids- and lipid-binding regions. Prediction of binding IDRs in protein sequences is gaining momentum in recent years. We survey 38 predictors of binding IDRs that target interactions with a diverse set of partners, such as peptides, proteins, RNA, DNA and lipids. We offer a historical perspective and highlight key events that fueled efforts to develop these methods. These tools rely on a diverse range of predictive architectures that include scoring functions, regular expressions, traditional and deep machine learning and meta-models. Recent efforts focus on the development of deep neural network-based architectures and extending coverage to RNA, DNA and lipid-binding IDRs. We analyze availability of these methods and show that providing implementations and webservers results in much higher rates of citations/use. We also make several recommendations to take advantage of modern deep network architectures, develop tools that bundle predictions of multiple and different types of binding IDRs, and work on algorithms that model structures of the resulting complexes.

MCWS-Transformers: Towards an Efficient Modeling of Protein Sequences via Multi Context-Window Based Scaled Self-Attention

This paper advances the self-attention mechanism in the standard transformer network specific to the modeling of the protein sequences. We introduce a novel context-window based scaled self-attention mechanism for processing protein sequences that is based on the notion of (i) local context and (ii) large contextual pattern. Both notions are essential to building a good representation for protein sequences. The proposed context-window based scaled self-attention mechanism is further used to build the multi context-window based scaled (MCWS) transformer network for the protein function prediction task at the protein sub-sequence level. Overall, the proposed MCWS transformer network produced improved predictive performances, outperforming existing state-of-the-art approaches by substantial margins. With respect to the standard transformer network, the proposed network produced improvements in F1-score of +2.30% and +2.08% on the biological process (BP) and molecular function (MF) datasets, respectively. The corresponding improvements over the state-of-the-art ProtVecGen-Plus+ProtVecGen-Ensemble approach are +3.38% (BP) and +2.86% (MF). Equally important, robust performances were obtained across protein sequences of different lengths.