Current Bioinformatics - Volume 19, Issue 4, 2024

Different diseases can be treated with various therapeutic agents. Drug discovery aims to find potential molecules for existing and emerging diseases. However, factors, such as increasing development cost, generic competition due to the patent expiry of several drugs, increase in conservative regulatory policies, and insufficient breakthrough innovations impairs the development of new drugs and the learning productivity of pharmaceutical industries. Drug repurposing is the process of finding new therapeutic applications for already approved, withdrawn from use, abandoned, and experimental drugs. Drug repurposing is another method that may partially overcome the hurdles related to drug discovery and hence appears to be a wise attempt. However, drug repurposing being not a standard regulatory process, leads to administrative concerns and problems. The drug repurposing also requires expensive, high-risk clinical trials to establish the safety and efficacy of the repurposed drug. Recent innovations in the field of bioinformatics can accelerate the new drug repurposing studies by identifying new targets of the existing drugs along with drug candidate screening and refinement. Recent advancements in the field of comprehensive high throughput data in genomics, epigenetics, chromosome architecture, transcriptomic, proteomics, and metabolomics may also contribute to the understanding of molecular mechanisms involved in drug-target interaction. The present review describes the current scenario in the field of drug repurposing along with the application of various bioinformatic tools for the identification of new targets for the existing drug.

Background: The prediction of drug-target interactions (DTIs) plays an essential role in drug discovery. Recently, deep learning methods have been widely applied in DTI prediction. However, most of the existing research does not fully utilize the molecular structures of drug compounds and the sequence structures of proteins, which makes these models unable to obtain precise and effective feature representations. Methods: In this study, we propose a novel deep learning framework combining transformer and graph neural networks for predicting DTIs. Our model utilizes graph convolutional neural networks to capture the global and local structure information of drugs, and convolutional neural networks are employed to capture the sequence feature of targets. In addition, the obtained drug and protein representations are input to multi-layer transformer encoders, respectively, to integrate their features and generate final representations. Results: The experiments on benchmark datasets demonstrated that our model outperforms previous graph-based and transformer-based methods, with 1.5% and 1.8% improvement in precision and 0.2% and 1.0% improvement in recall, respectively. The results indicate that the transformer encoders effectively extract feature information of both drug compounds and proteins. Conclusion: Overall, our proposed method validates the applicability of combining graph neural networks and transformer architecture in drug discovery, and due to the attention mechanisms, it can extract deep structure feature data of drugs and proteins.

Promoters are DNA fragments located near the transcription initiation site, they can be divided into strong promoter type and weak promoter type according to transcriptional activation and expression level. Identifying promoters and their strengths in DNA sequences is essential for understanding gene expression regulation. Therefore, it is crucial to further improve predictive quality of predictors for real-world application requirements. Here, we constructed the latest training dataset based on the RegalonDB website, where all the promoters in this dataset have been experimentally validated, and their sequence similarity is less than 85%. We used one-hot and nucleotide chemical property and density (NCPD) to represent DNA sequence samples. Additionally, we proposed an ensemble deep learning framework containing a multi-head attention module, long short-term memory present, and a convolutional neural network module. The results showed that iPSI(2L)-EDL outperformed other existing methods for both promoter prediction and identification of strong promoter type and weak promoter type, the AUC and MCC for the iPSI(2L)-EDL in identifying promoter were improved by 2.23% and 2.96% compared to that of PseDNC-DL on independent testing data, respectively, while the AUC and MCC for the iPSI(2L)- EDL were increased by 3.74% and 5.86% in predicting promoter strength type, respectively. The results of ablation experiments indicate that CNN plays a crucial role in recognizing promoters, the importance of different input positions and long-range dependency relationships among features are helpful for recognizing promoters. Furthermore, to make it easier for most experimental scientists to get the results they need, a userfriendly web server has been established and can be accessed at http://47.94.248.117/IPSW(2L)-EDL.

Background: LncRNA is not only involved in the regulation of the biological functions of protein-coding genes, but its dysfunction is also associated with the occurrence and progression of various diseases. Various studies have shown that an in-depth understanding of the mechanism of action of lncRNA is of great significance for disease treatment. However, traditional wet testing is time-consuming, laborious, expensive, and has many subjective factors which may affect the accuracy of the experiment. Objective: Most of the methods for predicting lncRNA-protein interaction (LPI) rely on a single feature, or there is noise in the feature. To solve this problem, we proposed a computational model, CSALPI based on a deep neural network. Methods: Firstly, this model utilizes cosine similarity to extract similarity features for lncRNAlncRNA and protein-protein, denoising similar features using the Sparse Autoencoder. Second, a neighbor enhancement autoencoder is employed to enforce neighboring nodes to be represented similarly by reconstructing the denoised features. Finally, a Light Gradient Boosting Machine classifier is used to predict potential LPIs. Results: To demonstrate the reliability of CSALPI, multiple evaluation metrics were used under a 5- fold cross-validation experiment, and excellent results were achieved. In the case study, the model successfully predicted 7 out of 10 disease-associated lncRNA and protein pairs. Conclusion: The CSALPI can be an effective complementary method for predicting potential LPIs from biological experiments.

Background: Predicting drug-target interaction (DTI) plays a crucial role in drug research and development. More and more researchers pay attention to the problem of developing more powerful prediction methods. Traditional DTI prediction methods are basically realized by biochemical experiments, which are time-consuming, risky, and costly. Nowadays, DTI prediction is often solved by using a single information source and a single model, or by combining some models, but the prediction results are still not accurate enough. Objective: The study aimed to utilize existing data and machine learning models to integrate heterogeneous data sources and different models, further improving the accuracy of DTI prediction. Methods: This paper has proposed a novel prediction method based on reinforcement learning, called QLDTI (predicting drug-target interaction based on Q-learning), which can be mainly divided into two parts: data fusion and model fusion. Firstly, it fuses the drug and target similarity matrices calculated by different calculation methods through Q-learning. Secondly, the new similarity matrices are inputted into five models, NRLMF, CMF, BLM-NII, NetLapRLS, and WNN-GIP, for further training. Then, all sub-model weights are continuously optimized again by Q-learning, which can be used to linearly weight all sub-model prediction results to output the final prediction result. Results: QLDTI achieved AUC accuracy of 99.04%, 99.12%, 98.28%, and 98.35% on E, NR, IC, and GPCR datasets, respectively. Compared to the existing five models NRLMF, CMF, BLM-NII, NetLapRLS, and WNN-GIP, the QLDTI method has achieved better results on four benchmark datasets of E, NR, IC, and GPCR. Conclusion: Data fusion and model fusion have been proven effective for DTI prediction, further improving the prediction accuracy of DTI.

Background: Pathological myopia, a severe variant of myopia, extends beyond the typical refractive error associated with nearsightedness. While the condition has a strong genetic component, the intricate mechanisms of inheritance remain elusive. Some genes have been associated with the development of pathological myopia, but their exact roles are not fully understood. Objective: This study aimed to identify novel genes associated with pathological myopia. Methods: Our study leveraged DisGeNET to identify 184 genes linked with high myopia and 39 genes related to degenerative myopia. To uncover additional pathological myopia-associated genes, we employed the random walk with restart algorithm to investigate the protein-protein interactions network. We used the previously identified 184 high myopia and 39 degenerative myopia genes as seed nodes. Results: Through subsequent screening tests, we discarded genes with weak associations, yielding 103 new genes for high myopia and 33 for degenerative myopia. Conclusion: We confirmed the association of certain genes, including six genes that were confirmed to be associated with both high and degenerative myopia. The newly discovered genes are helpful to uncover and understand the pathogenesis of myopia.

Background: Interactions of RNA and DNA with proteins are crucial for elucidating intracellular processes in living organisms, diagnosing disorders, designing aptamer drugs, and other applications. Therefore, investigating the relationships between these macromolecules is essential to life science research. Methods: This study proposes an online network provider tool (NaProGraph) that offers an intuitive and user-friendly interface for studying interactions between nucleic acids (NA) and proteins. NaPro- Graph utilizes a comprehensive and curated dataset encompassing nearly all interacting macromolecules in the Research Collaboratory for Structural Bioinformatics (RCSB) Protein Data Bank (PDB). Results: Researchers can employ this online tool to focus on a specific portion of the PDB, investigate its associated relationships, and visualize and extract pertinent information. This tool provides insights into the frequency of atoms and residues between proteins and nucleic acids (NAs) and the similarity of the macromolecules' primary structures. Conclusion: Furthermore, the functional similarity of proteins can be inferred using protein families and clans from Pfam.

Introduction: Transcription factors are of great interest in biotechnology due to their key role in the regulation of gene expression. One of the most important transcription factors in gramnegative bacteria is Fur, a global regulator studied as a therapeutic target for the design of antibacterial agents. Its DNA-binding domain, which contains a helix-turn-helix motif, is one of its most relevant features. Methods: In this study, we evaluated several machine learning algorithms for the prediction of DNA-binding sites based on proteins from the Fur superfamily and other helix-turn-helix transcription factors, including Support-Vector Machines (SVM), Random Forest (RF), Decision Trees (DT), and Naive Bayes (NB). We also tested the efficacy of using several molecular descriptors derived from the amino acid sequence and the structure of the protein fragments that bind the DNA. A feature selection procedure was employed to select fewer descriptors in each case by maintaining a good classification performance. Results: The best results were obtained with the SVM model using twelve sequence-derived attributes and the DT model using nine structure-derived features, achieving 82% and 76% accuracy, respectively. Conclusion: The performance obtained indicates that the descriptors we used are relevant for predicting DNA-binding sites since they can discriminate between binding and non-binding regions of a protein.

Aims: Bulk and single-cell RNA sequencing data were analyzed to explore the association of stemness phenotype with dysfunctional anti-tumor immunity and its impact on clinical outcomes of primary and relapse HCC. Background: The stemness phenotype is gradually acquired during cancer progression; however, it remains unclear the effect of stemness phenotype on recurrence and clinical outcomes in hepatocellular carcinoma (HCC). Methods: The stemness index (mRNAsi) calculated by a one-class logistic regression algorithm in multiple HCC cohorts was defined as the stemness phenotype of the patient. Using single-cell profiling in primary or early-relapse HCC, cell stemness phenotypes were evaluated by developmental potential. Differential analysis of stemness phenotype, gene expression and interactions between primary and recurrent samples revealed the underlying immune evasion mechanisms. Results: A strong correlation was discovered between mRNAsi and clinical outcomes in patient with HCC. The high and low mRNAsi groups had distinct tumor immune microenvironments. Cellular stemness phenotype varied by cell type. Moreover, compared with primary tumors, early-relapse tumors had increased stemness of dendritic cells and tumor cells and reduced stemness of T cells and B cells. Moreover, in relapse tumors, CD8+ T cells displayed a low stemness state, with a high exhausted state, unlike the high stemness state observed in primary HCC. Conclusions: The comprehensive characterization of the HCC stemness phenotype provides insights into the clinical outcomes and immune escape mechanisms associated with recurrence.