A High Efficient Biological Language Model for Predicting Protein⁻Protein Interactions.

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/.

Protein-protein interactions (PPIs) are crucial to most biochemical processes in human beings. Although many human PPIs have been identified by experiments, the number is still limited compared to the available protein sequences of human organisms. Recently, many computational methods have been proposed to facilitate the recognition of novel human PPIs. However the existing methods only concentrated on the information of individual PPI, while the systematic characteristic of protein-protein interaction networks (PINs) was ignored. In this study, a new method was proposed by combining the global information of PINs and protein sequence information. Random forest (RF) algorithm was implemented to develop the prediction model, and a high accuracy of 91.88% was obtained. Furthermore, the RF model was tested using three independent datasets with good performances, suggesting that our method is a useful tool for identification of PPIs and investigation into PINs as well.

Protein-protein interactions (PPIs) play a crucial role in the life cycles of living cells. Thus, it is important to understand the underlying mechanisms of PPIs. Although many high-throughput technologies have generated large amounts of PPI data in different organisms, the experiments for detecting PPIs are still costly and time-consuming. Therefore, novel computational methods are urgently needed for predicting PPIs. For this reason, developing a new computational method for predicting PPIs is drawing more and more attention. In this study, we proposed a novel computational method based on texture feature of protein sequence for predicting PPIs. Especially, the Gabor feature is used to extract texture feature and protein evolutionary information from Position-Specific Scoring Matrix, which is generated by Position-Specific Iterated Basic Local Alignment Search Tool. Then, random forest–based classifiers are used to infer the protein interactions. When performed on PPI data sets of yeast, human, and Helicobacter pylori, we obtained good results with average accuracies of 92.10%, 97.03%, and 86.45%, respectively. To better evaluate the proposed method, we compared Gabor feature, Discrete Cosine Transform, and Local Phase Quantization. Our results show that the proposed method is both feasible and stable and the Gabor feature descriptor is reliable in extracting protein sequence information. Furthermore, additional experiments have been conducted to predict PPIs of other 4 species data sets. The promising results indicate that our proposed method is both powerful and robust.

The Impact of Gene Expression Regulation on Evolution of Extracellular Signaling Pathways

Structural genomics provides an important approach for characterizing and understanding systems biology. As a step toward better integrating protein three-dimensional (3D) structural information in cancer systems biology, we have constructed a Human Cancer Pathway Protein Interaction Network (HCPIN) by analysis of several classical cancer-associated signaling pathways and their physical protein-protein interactions. Many well known cancer-associated proteins play central roles as "hubs" or "bottlenecks" in the HCPIN. At least half of HCPIN proteins are either directly associated with or interact with multiple signaling pathways. Although some 45% of residues in these proteins are in sequence segments that meet criteria sufficient for approximate homology modeling (Basic Local Alignment Search Tool (BLAST) E-value <10(-6)), only approximately 20% of residues in these proteins are structurally covered using high accuracy homology modeling criteria (i.e. BLAST E-value <10(-6) and at least 80% sequence identity) or by actual experimental structures. The HCPIN Website provides a comprehensive description of this biomedically important multipathway network together with experimental and homology models of HCPIN proteins useful for cancer biology research. To complement and enrich cancer systems biology, the Northeast Structural Genomics Consortium is targeting >1000 human proteins and protein domains from the HCPIN for sample production and 3D structure determination. The long range goal of this effort is to provide a comprehensive 3D structure-function database for human cancer-associated proteins and protein complexes in the context of their interaction networks. The network-based target selection (BioNet) approach described here is an example of a general strategy for targeting co-functioning proteins by structural genomics projects.

Zea mays (maize) is one of the most vital crops which are grown widely in the world. To understand the molecular structures and functions of maize, the identification of protein-protein interaction (PPI) is very important. PPI identification by wet lab experiments is time-consuming, expensive and laborious. These days in silico methods that accurately predict potential PPIs based on protein sequence information are highly demanded. Research on PPI prediction in maize is currently very limited, and no dedicated bioinformatics schemes are available. In this work, we proposed a novel approach, termed SIPMA (Systematic Identification of PPI in Maize using Autocorrelation). A machine learning random forest classifier was trained with autocorrelation features to build the prediction model. The SIPMA, which was tested by the experimentally verified PPI dataset of maize, yielded a prediction accuracy of 0.899 when the specificity was 0.969 on the training set. The SIPMA achieved promising performances on the test datasets. Compared with different sequence-based encoding and statistical learning methods, the SIPMA was a powerful computational resource for identifying PPIs in maize.

Convolutional neural networks (CNNs) have been used to extract information from various datasets of different dimensions. This approach has led to accurate interpretations in several subfields of biological research, like pharmacogenomics, addressing issues previously faced by other computational methods. With the rising attention for personalized and precision medicine, scientists and clinicians have now turned to artificial intelligence systems to provide them with solutions for therapeutics development. CNNs have already provided valuable insights into biological data transformation. Due to the rise of interest in precision and personalized medicine, in this review, we have provided a brief overview of the possibilities of implementing CNNs as an effective tool for analyzing one-dimensional biological data, such as nucleotide and protein sequences, as well as small molecular data, e.g., simplified molecular-input line-entry specification, InChI, binary fingerprints, etc., to categorize the models based on their objective and also highlight various challenges. The review is organized into specific research domains that participate in pharmacogenomics for a more comprehensive understanding. Furthermore, the future intentions of deep learning are outlined.

The subcellular location of a protein provides valuable insights to bioinformaticians in terms of drug designs and discovery, genomics, and various other aspects of medical research. Experimental methods for protein subcellular localization determination are time-consuming and expensive, whereas computational methods, if accurate, would represent a much more efficient alternative. This article introduces an ab initio protein subcellular localization predictor based on an ensemble of Deep N-to-1 Convolutional Neural Networks. Our predictor is trained and tested on strict redundancy-reduced datasets and achieves 63% accuracy for the diverse number of classes. This predictor is a step towards bridging the gap between a protein sequence and the protein's function. It can potentially provide information about protein-protein interaction to facilitate drug design and processes like vaccine production that are essential to disease prevention.

In the postgenomic era, the large-scale data, such as genome sequences, mRNA sequences, and protein sequences, increase rapidly. It is desired to develop the computational approaches that can derive and analyze useful information from them to promote the development of biomedicine and drug design. Meanwhile, in order to understand how proteinprotein interactions and other complex interactions in a living system get integrated in complex nonlinear networks and regulate cell function, a new discipline, called "Systems Biology", is created.

MotivationProtein function prediction is important for drug development and disease treatment. Recently, deep learning methods have leveraged protein sequence and structural information, achieving remarkable progress in the field of protein function prediction. However, existing methods ignore the complex multimodal interaction information between sequence and structural features. Since protein sequence and structural information reveal the functional characteristics of proteins from different perspectives, it is challenging to effectively fuse the information from these two modalities to portray protein functions more comprehensively. In addition, current methods have difficulty in effectively capturing long-range dependencies and global contextual information in protein sequences during feature extraction, thus limiting the ability of the model to recognize critical functional residues.ResultsIn this study, we propose a novel framework termed Multi-stage Attention-based Extraction and Fusion model for GO prediction (MAEF-GO) based on a multistage attention mechanism to predict protein functions. MAEF-GO innovatively integrates the graph convolutional network and the graph attention network to extract protein structural features. To address the issue of modeling long-range dependencies within protein sequences, we introduce a frequency-domain attention mechanism capable of extracting global contextual relationships. Additionally, a cross-attention module is implemented to facilitate interactive fusion between protein sequence and structural modalities. Experimental evaluations demonstrate that MAEF-GO achieves superior performance compared to several state-of-the-art baseline models across standard benchmarks. Furthermore, analysis of the cross-attention weight distributions demonstrates MAEF-GO’s interpretability. It can effectively identify critical functional residues of proteins.Availability and implementationThe MAEF-GO source code can be found at https://github.com/nebstudio/MAEF-GO, an archived snapshot of the code used in this study is also available via Zenodo at https://doi.org/10.5281/zenodo.15422392.

Co-evolution has an important function in the evolution of species and it is clearly manifested in certain scenarios such as host–parasite and predator–prey interactions, symbiosis and mutualism. The extrapolation of the concepts and methodologies developed for the study of species co-evolution at the molecular level has prompted the development of a variety of computational methods able to predict protein interactions through the characteristics of co-evolution. Particularly successful have been those methods that predict interactions at the genomic level based on the detection of pairs of protein families with similar evolutionary histories (similarity of phylogenetic trees: mirrortree). Future advances in this field will require a better understanding of the molecular basis of the co-evolution of protein families. Thus, it will be important to decipher the molecular mechanisms underlying the similarity observed in phylogenetic trees of interacting proteins, distinguishing direct specific molecular interactions from other general functional constraints. In particular, it will be important to separate the effects of physical interactions within protein complexes (‘co-adaptation') from other forces that, in a less specific way, can also create general patterns of co-evolution.

Prediction of protein-protein interaction (PPI) is one of the most challenging problems in biology. Although great progress has been devoted to the development of methodology for predicting PPIs and PIN using machine learning methods, the problem is still far from being solved since the application of most existing methods is limited. In this study, we propose a method for PPI prediction based on amino acids differences between pairs of protein sequences. 10-fold cross-validation tests based on human PPI datasets with balanced positive-to-negative ratios indicate that it performs comparably well. Therefore, our finding suggests that amino acids differences of interacting protein pairs are relevant to the prediction of PPIs and hence provide important information on sequence-based encoding schemes.

The high-throughput technologies have led to vast amounts of protein-protein interaction (PPI) data, and a number of approaches based on PPI networks have been proposed for protein function prediction. However, these approaches do not work well if there is not enough PPI information. To address this issue, we propose a novel collective classification based approach that combines protein sequence information and PPI information to improve the prediction performance. We first reconstruct a PPI network by adding a number of computed edges based on protein sequence similarity, and then apply a collective classification algorithm to predict protein function based on the new PPI network. Experiments over two real datasets demonstrate that our approach outperforms most of existing approaches across a series of label situations, especially in sparsely-labeled networks where the existing approaches fail because of PPI information inadequacy. Experimental results also validate the robustness of our approach to the number of labeled proteins in PPI networks.

Protein-protein interactions (PPIs) play key roles in life processes, such as signal transduction, transcription regulations, and immune response, etc. Identification of PPIs enables better understanding of the functional networks within a cell. Common experimental methods for identifying PPIs are time consuming and expensive. However, recent developments in computational approaches for inferring PPIs from protein sequences based on coevolution theory avoid these problems. In the coevolution theory model, interacted proteins may show coevolutionary mutations and have similar phylogenetic trees. The existing coevolution methods depend on multiple sequence alignments (MSA); however, the MSA-based coevolution methods often produce high false positive interactions. In this paper, we present a computational method using an alignment-free approach to accurately detect PPIs and reduce false positives. In the method, protein sequences are numerically represented by biochemical properties of amino acids, which reflect the structural and functional differences of proteins. Fourier transform is applied to the numerical representation of protein sequences to capture the dissimilarities of protein sequences in biophysical context. The method is assessed for predicting PPIs in Ebola virus. The results indicate strong coevolution between the protein pairs (NP-VP24, NP-VP30, NP-VP40, VP24-VP30, VP24-VP40, and VP30-VP40). The method is also validated for PPIs in influenza and E.coli genomes. Since our method can reduce false positive and increase the specificity of PPI prediction, it offers an effective tool to understand mechanisms of disease pathogens and find potential targets for drug design. The Python programs in this study are available to public at URL (https://github.com/cyinbox/PPI).

Protein interactions shape proteome function and thus biology. Identification of protein interactions is a major goal in molecular biology, but biochemical methods, although improving, remain limited in coverage and accuracy. Whereas computational predictions can guide biochemical experiments, low validation rates of predictions remain a major limitation. Here, we investigated computational methods in the prediction of a specific type of interaction, the inhibitory interactions between proteases and their inhibitors. Proteases generate thousands of proteoforms that dynamically shape the functional state of proteomes. Despite the important regulatory role of proteases, knowledge of their inhibitors remains largely incomplete with the vast majority of proteases lacking an annotated inhibitor. To link inhibitors to their target proteases on a large scale, we applied computational methods to predict inhibitory interactions between proteases and their inhibitors based on complementary data, including coexpression, phylogenetic similarity, structural information, co-annotation, and colocalization, and also surveyed general protein interaction networks for potential inhibitory interactions. In testing nine predicted interactions biochemically, we validated the inhibition of kallikrein 5 by serpin B12. Despite the use of a wide array of complementary data, we found a high false positive rate of computational predictions in biochemical follow-up. Based on a protease-specific definition of true negatives derived from the biochemical classification of proteases and inhibitors, we analyzed prediction accuracy of individual features, thereby we identified feature-specific limitations, which also affected general protein interaction prediction methods. Interestingly, proteases were often not coexpressed with most of their functional inhibitors, contrary to what is commonly assumed and extrapolated predominantly from cell culture experiments. Predictions of inhibitory interactions were indeed more challenging than predictions of nonproteolytic and noninhibitory interactions. In summary, we describe a novel and well-defined but difficult protein interaction prediction task and thereby highlight limitations of computational interaction prediction methods.