Machine learning on protein-protein interaction prediction: models, challenges and trends

Structure-independent machine-learning predictions of the CDK12 interactome

Cyclin-dependent kinase 12 (CDK12) is a critical regulatory protein involved in transcription and DNA repair processes. Dysregulation of CDK12 has been implicated in various diseases, including cancer. Understanding the CDK12 interactome is pivotal for elucidating its functional roles and potential therapeutic targets. Traditional methods for interactome prediction often rely on protein structure information, limiting applicability to CDK12 characterized by partly disordered terminal C region. In this study, we present a structure-independent machine-learning model that utilizes proteins' sequence and functional data to predict the CDK12 interactome. This approach is motivated by the disordered character of the CDK12 C-terminal region mitigating a structure-driven search for binding partners. Our approach incorporates multiple data sources, including protein-protein interaction networks, functional annotations, and sequence-based features, to construct a comprehensive CDK12 interactome prediction model. The ability to predict CDK12 interactions without relying on structural information is a significant advancement, as many potential interaction partners may lack crystallographic data. In conclusion, our structure-independent machine-learning model presents a powerful tool for predicting the CDK12 interactome and holds promise in advancing our understanding of CDK12 biology, identifying potential therapeutic targets, and facilitating precision-medicine approaches for CDK12-associated diseases.

TUnA: an uncertainty-aware transformer model for sequence-based protein-protein interaction prediction

Protein-protein interactions (PPIs) are important for many biological processes, but predicting them from sequence data remains challenging. Existing deep learning models often cannot generalize to proteins not present in the training set and do not provide uncertainty estimates for their predictions. To address these limitations, we present TUnA, a Transformer-based uncertainty-aware model for PPI prediction. TUnA uses ESM-2 embeddings with Transformer encoders and incorporates a Spectral-normalized Neural Gaussian Process. TUnA achieves state-of-the-art performance and, importantly, evaluates uncertainty for unseen sequences. We demonstrate that TUnA's uncertainty estimates can effectively identify the most reliable predictions, significantly reducing false positives. This capability is crucial in bridging the gap between computational predictions and experimental validation.

SENSE-PPI reconstructs interactomes within, across, and between species at the genome scale

Ab initio computational reconstructions of protein-protein interaction (PPI) networks will provide invaluable insights into cellular systems, enabling the discovery of novel molecular interactions and elucidating biological mechanisms within and between organisms. Leveraging the latest generation protein language models and recurrent neural networks, we present SENSE-PPI, a sequence-based deep learning model that efficiently reconstructs ab initio PPIs, distinguishing partners among tens of thousands of proteins and identifying specific interactions within functionally similar proteins. SENSE-PPI demonstrates high accuracy, limited training requirements, and versatility in cross-species predictions, even with non-model organisms and human-virus interactions. Its performance decreases for phylogenetically more distant model and non-model organisms, but signal alteration is very slow. In this regard, it demonstrates the important role of parameters in protein language models. SENSE-PPI is very fast and can test 10,000 proteins against themselves in a matter of hours, enabling the reconstruction of genome-wide proteomes.

Prediction of protein secondary structure by the improved TCN-BiLSTM-MHA model with knowledge distillation

Secondary structure prediction is a key step in understanding protein function and biological properties and is highly important in the fields of new drug development, disease treatment, bioengineering, etc. Accurately predicting the secondary structure of proteins helps to reveal how proteins are folded and how they function in cells. The application of deep learning models in protein structure prediction is particularly important because of their ability to process complex sequence information and extract meaningful patterns and features, thus significantly improving the accuracy and efficiency of prediction. In this study, a combined model integrating an improved temporal convolutional network (TCN), bidirectional long short-term memory (BiLSTM), and a multi-head attention (MHA) mechanism is proposed to enhance the accuracy of protein prediction in both eight-state and three-state structures. One-hot encoding features and word vector representations of physicochemical properties are incorporated. A significant emphasis is placed on knowledge distillation techniques utilizing the ProtT5 pretrained model, leading to performance improvements. The improved TCN, achieved through multiscale fusion and bidirectional operations, allows for better extraction of amino acid sequence features than traditional TCN models. The model demonstrated excellent prediction performance on multiple datasets. For the TS115, CB513 and PDB (2018-2020) datasets, the prediction accuracy of the eight-state structure of the six datasets in this paper reached 88.2%, 84.9%, and 95.3%, respectively, and the prediction accuracy of the three-state structure reached 91.3%, 90.3%, and 96.8%, respectively. This study not only improves the accuracy of protein secondary structure prediction but also provides an important tool for understanding protein structure and function, which is particularly applicable to resource-constrained contexts and provides a valuable tool for understanding protein structure and function.

MGPPI: multiscale graph neural networks for explainable protein-protein interaction prediction

Protein-Protein Interactions (PPIs) involves in various biological processes, which are of significant importance in cancer diagnosis and drug development. Computational based PPI prediction methods are more preferred due to their low cost and high accuracy. However, existing protein structure based methods are insufficient in the extraction of protein structural information. Furthermore, most methods are less interpretable, which hinder their practical application in the biomedical field. In this paper, we propose MGPPI, which is a Multiscale graph convolutional neural network model for PPI prediction. By incorporating multiscale module into the Graph Neural Network (GNN) and constructing multi convolutional layers, MGPPI can effectively capture both local and global protein structure information. For model interpretability, we introduce a novel visual explanation method named Gradient Weighted interaction Activation Mapping (Grad-WAM), which can highlight key binding residue sites. We evaluate the performance of MGPPI by comparing with state-of-the-arts methods on various datasets. Results shows that MGPPI outperforms other methods significantly and exhibits strong generalization capabilities on the multi-species dataset. As a practical case study, we predicted the binding affinity between the spike (S) protein of SARS-COV-2 and the human ACE2 receptor protein, and successfully identified key binding sites with known binding functions. Key binding sites mutation in PPIs can affect cancer patient survival statues. Therefore, we further verified Grad-WAM highlighted residue sites in separating patients survival groups in several different cancer type datasets. According to our results, some of the highlighted residues can be used as biomarkers in predicting patients survival probability. All these results together demonstrate the high accuracy and practical application value of MGPPI. Our method not only addresses the limitations of existing approaches but also can assists researchers in identifying crucial drug targets and help guide personalized cancer treatment.

Reconstruction of Eriocheir sinensis Protein-Protein Interaction Network Based on DGO-SVM Method

Eriocheir sinensis is an economically important aquatic animal. Its regulatory mechanisms underlying many biological processes are still vague due to the lack of systematic analysis tools. The protein-protein interaction network (PIN) is an important tool for the systematic analysis of regulatory mechanisms. In this work, a novel machine learning method, DGO-SVM, was applied to predict the protein-protein interaction (PPI) in E. sinensis, and its PIN was reconstructed. With the domain, biological process, molecular functions and subcellular locations of proteins as the features, DGO-SVM showed excellent performance in Bombyx mori, humans and five aquatic crustaceans, with 92-96% accuracy. With DGO-SVM, the PIN of E. sinensis was reconstructed, containing 14,703 proteins and 7,243,597 interactions, in which 35,604 interactions were associated with 566 novel proteins mainly involved in the response to exogenous stimuli, cellular macromolecular metabolism and regulation. The DGO-SVM demonstrated that the biological process, molecular functions and subcellular locations of proteins are significant factors for the precise prediction of PPIs. We reconstructed the largest PIN for E. sinensis, which provides a systematic tool for the regulatory mechanism analysis. Furthermore, the novel-protein-related PPIs in the PIN may provide important clues for the mechanism analysis of the underlying specific physiological processes in E. sinensis.

Co-training based prediction of multi-label protein-protein interactions

Prediction of protein-protein interaction (PPI) types enhances the comprehension of the underlying structural characteristics and functions of proteins, which gives rise to a multi-label classification problem. The nominal features describe the physicochemical characteristics of proteins directly, establishing a more robust correlation with the interaction types between proteins than ordered features. Motivated by this, we propose a multi-label PPI prediction model referred to as CoMPPI (Co-training based Multi-Label prediction of Protein-Protein Interaction). This approach aims to maximize the utility of both ordered and nominal features extracted from protein sequences. Specifically, CoMPPI incorporates graph convolutional network (GCN) and 1D convolution operation to process the complementary subsets of features individually, leveraging both local and contextualized information in a more efficient way. In addition, two multi-type PPI datasets were constructed to eliminate the duplication in previous datasets. We compare the performance of CoMPPI with three state-of-the-art methods on three datasets partitioned using distinct schemes (Breadth-first search, Depth-first search, and Random), CoMPPI consistently outperforms the other methods across all cases, demonstrating improvements ranging from 3.81% to 32.40% in Micro-F1. The subsequent ablation experiment confirms the efficacy of employing the co-training framework for multi-label PPI prediction, indicating promising avenues for future advancements in this domain.

Immunological parameters of maternal peripheral blood as predictors of future pregnancy outcomes in patients with unexplained recurrent pregnancy loss

INTRODUCTION

Unexplained recurrent pregnancy loss (URPL), affecting approximately 1%-5% of women, exhibits a strong association with various maternal factors, particularly immune disorders. However, accurately predicting pregnancy outcomes based on the complex interactions and synergistic effects of various immune parameters without an automated algorithm remains challenging.

MATERIAL AND METHODS

In this historical cohort study, we analyzed the medical records of URPL patients treated at Xiangya Hospital, Changsha, China, between January 2020 and October 2022. The primary outcomes included clinical pregnancy and miscarriage. Predictors included complement, autoantibodies, peripheral lymphocytes, immunoglobulins, thromboelastography findings, and serum lipids. Least absolute shrinkage and selection operator (LASSO) analysis and logistic regression analysis was performed for model development. The model's performance, discriminatory, and clinical applicability were assessed using area under the curve (AUC), calibration curve, and decision curve analysis, respectively. Additionally, models were visualized by constructing dynamic and static nomograms.

RESULTS

In total, 502 patients with URPL were enrolled, of whom 291 (58%) achieved clinical pregnancy and 211 (42%) experienced miscarriage. Notable differences in complement, peripheral lymphocytes, and serum lipids were observed between the two outcome groups. Moreover, URPL patients with elevated peripheral NK cells (absolute counts and proportion), decreased complement levels, and dyslipidemia demonstrated a significantly increased risk of miscarriage. Four models were developed in this study, of which Model 2 demonstrated superior performance with only seven predictors, achieving an AUC of 0.96 (95% CI: 0.93-0.99) and an accuracy of 0.92. A web-based platform was established to visually present model 2 and to facilitate its utilization by clinicians in outpatient settings (available from: https://yingrongli.shinyapps.io/liyingrong/).

CONCLUSIONS

Our findings suggest that the implementation of such prediction models could serve as valuable tools for providing comprehensive information and facilitating clinicians in their decision-making processes.

Integration of hybrid and self-correction method improves the quality of long-read sequencing data

Third-generation sequencing (TGS) technologies have revolutionized genome science in the past decade. However, the long-read data produced by TGS platforms suffer from a much higher error rate than that of the previous technologies, thus complicating the downstream analysis. Several error correction tools for long-read data have been developed; these tools can be categorized into hybrid and self-correction tools. So far, these two types of tools are separately investigated, and their interplay remains understudied. Here, we integrate hybrid and self-correction methods for high-quality error correction. Our procedure leverages the inter-similarity between long-read data and high-accuracy information from short reads. We compare the performance of our method and state-of-the-art error correction tools on Escherichia coli and Arabidopsis thaliana datasets. The result shows that the integration approach outperformed the existing error correction methods and holds promise for improving the quality of downstream analyses in genomic research.

Leveraging machine learning models for peptide-protein interaction prediction

Peptides play a pivotal role in a wide range of biological activities through participating in up to 40% protein-protein interactions in cellular processes. They also demonstrate remarkable specificity and efficacy, making them promising candidates for drug development. However, predicting peptide-protein complexes by traditional computational approaches, such as docking and molecular dynamics simulations, still remains a challenge due to high computational cost, flexible nature of peptides, and limited structural information of peptide-protein complexes. In recent years, the surge of available biological data has given rise to the development of an increasing number of machine learning models for predicting peptide-protein interactions. These models offer efficient solutions to address the challenges associated with traditional computational approaches. Furthermore, they offer enhanced accuracy, robustness, and interpretability in their predictive outcomes. This review presents a comprehensive overview of machine learning and deep learning models that have emerged in recent years for the prediction of peptide-protein interactions.

Machine Learning Empowering Drug Discovery: Applications, Opportunities and Challenges

Drug discovery plays a critical role in advancing human health by developing new medications and treatments to combat diseases. How to accelerate the pace and reduce the costs of new drug discovery has long been a key concern for the pharmaceutical industry. Fortunately, by leveraging advanced algorithms, computational power and biological big data, artificial intelligence (AI) technology, especially machine learning (ML), holds the promise of making the hunt for new drugs more efficient. Recently, the Transformer-based models that have achieved revolutionary breakthroughs in natural language processing have sparked a new era of their applications in drug discovery. Herein, we introduce the latest applications of ML in drug discovery, highlight the potential of advanced Transformer-based ML models, and discuss the future prospects and challenges in the field.

scCAN: Clustering With Adaptive Neighbor-Based Imputation Method for Single-Cell RNA-Seq Data

Single-cell RNA sequencing (scRNA-seq) is widely used to study cellular heterogeneity in different samples. However, due to technical deficiencies, dropout events often result in zero gene expression values in the gene expression matrix. In this paper, we propose a new imputation method called scCAN, based on adaptive neighborhood clustering, to estimate the zero value of dropouts. Our method continuously updates cell-cell similarity information by simultaneously learning similarity relationships, clustering structures, and imposing new rank constraints on the Laplacian matrix of the similarity matrix, improving the imputation of dropout zero values. To evaluate the performance of this method, we used four simulated and eight real scRNA-seq data for downstream analyses, including cell clustering, recovered gene expression, and reconstructed cell trajectories. Our method improves the performance of the downstream analysis and is better than other imputation methods.

Oligomerization of Monoamine Transporters

Transporters of the monoamine transporter (MAT) family regulate the uptake of important neurotransmitters like dopamine, serotonin, and norepinephrine. The MAT family functions using the electrochemical gradient of ions across the membrane and comprises three transporters, dopamine transporter (DAT), serotonin transporter (SERT), and norepinephrine transporter (NET). MAT transporters have been observed to exist in monomeric states to higher-order oligomeric states. Structural features, allosteric modulation, and lipid environment regulate the oligomerization of MAT transporters. NET and SERT oligomerization are regulated by levels of PIP2 present in the membrane. The kink present in TM12 in the MAT family is crucial for dimer interface formation. Allosteric modulation in the dimer interface hinders dimer formation. Oligomerization also influences the transporters' function, trafficking, and regulation. This chapter will focus on recent studies on monoamine transporters and discuss the factors affecting their oligomerization and its impact on their function.

Prediction of multi-relational drug-gene interaction via Dynamic hyperGraph Contrastive Learning

Drug-gene interaction prediction occupies a crucial position in various areas of drug discovery, such as drug repurposing, lead discovery and off-target detection. Previous studies show good performance, but they are limited to exploring the binding interactions and ignoring the other interaction relationships. Graph neural networks have emerged as promising approaches owing to their powerful capability of modeling correlations under drug-gene bipartite graphs. Despite the widespread adoption of graph neural network-based methods, many of them experience performance degradation in situations where high-quality and sufficient training data are unavailable. Unfortunately, in practical drug discovery scenarios, interaction data are often sparse and noisy, which may lead to unsatisfactory results. To undertake the above challenges, we propose a novel Dynamic hyperGraph Contrastive Learning (DGCL) framework that exploits local and global relationships between drugs and genes. Specifically, graph convolutions are adopted to extract explicit local relations among drugs and genes. Meanwhile, the cooperation of dynamic hypergraph structure learning and hypergraph message passing enables the model to aggregate information in a global region. With flexible global-level messages, a self-augmented contrastive learning component is designed to constrain hypergraph structure learning and enhance the discrimination of drug/gene representations. Experiments conducted on three datasets show that DGCL is superior to eight state-of-the-art methods and notably gains a 7.6% performance improvement on the DGIdb dataset. Further analyses verify the robustness of DGCL for alleviating data sparsity and over-smoothing issues.

Key protein identification by integrating protein complex information and multi-biological features

Identifying key proteins based on protein-protein interaction networks has emerged as a prominent area of research in bioinformatics. However, current methods exhibit certain limitations, such as the omission of subcellular localization information and the disregard for the impact of topological structure noise on the reliability of key protein identification. Moreover, the influence of proteins outside a complex but interacting with proteins inside the complex on complex participation tends to be overlooked. Addressing these shortcomings, this paper presents a novel method for key protein identification that integrates protein complex information with multiple biological features. This approach offers a comprehensive evaluation of protein importance by considering subcellular localization centrality, topological centrality weighted by gene ontology (GO) similarity and complex participation centrality. Experimental results, including traditional statistical metrics, jackknife methodology metric and key protein overlap or difference, demonstrate that the proposed method not only achieves higher accuracy in identifying key proteins compared to nine classical methods but also exhibits robustness across diverse protein-protein interaction networks.

Accelerating the Screening of Small Peptide Ligands by Combining Peptide-Protein Docking and Machine Learning

This research introduces a novel pipeline that couples machine learning (ML), and molecular docking for accelerating the process of small peptide ligand screening through the prediction of peptide-protein docking. Eight ML algorithms were analyzed for their potential. Notably, Light Gradient Boosting Machine (LightGBM), despite having comparable F1-score and accuracy to its counterparts, showcased superior computational efficiency. LightGBM was used to classify peptide-protein docking performance of the entire tetrapeptide library of 160,000 peptide ligands against four viral envelope proteins. The library was classified into two groups, 'better performers' and 'worse performers'. By training the LightGBM algorithm on just 1% of the tetrapeptide library, we successfully classified the remaining 99%with an accuracy range of 0.81-0.85 and an F1-score between 0.58-0.67. Three different molecular docking software were used to prove that the process is not software dependent. With an adjustable probability threshold (from 0.5 to 0.95), the process could be accelerated by a factor of at least 10-fold and still get 90-95% concurrence with the method without ML. This study validates the efficiency of machine learning coupled to molecular docking in rapidly identifying top peptides without relying on high-performance computing power, making it an effective tool for screening potential bioactive compounds.