Muhammad Arif

Protein post-translational modifications (PTMs) are critical for regulating protein function and are closely linked to disease mechanisms. In-depth research and precise prediction of PTMs are vital for understanding life mechanisms, screening disease biomarkers, and identifying drug targets. Artificial intelligence (AI) approaches for PTM site prediction offer complementary advantages to traditional experimental methods, providing high-throughput and cost-effective screening that can prioritize candidate sites for further validation. This paper reviews advances in PTM site prediction since 2012, focusing on machine learning and deep learning techniques. It analyzes more than 500 relevant studies and categorizes 36 types of PTMs. Additionally, the paper briefly outlines core contents such as database resources related to PTMs, commonly used feature extraction methods, and major classification algorithms. In addition, 36 representative recent studies on PTMs have been carefully selected for in-depth analysis. The findings indicate that current machine learning-based PTM research employs multivariate feature extraction and construct composite models to enhance prediction performance. Finally, keyword visualization using CiteSpace identifies emerging research hotspots and future directions for PTM site prediction.

Liquid–liquid phase separation (LLPS) is a biomolecular process that underpins the formation of membrane-less organelles within living cells. This phenomenon, along with the resulting condensate bodies, is increasingly recognized for its critical roles in various biological processes, such as ribonucleic acid (RNA) metabolism, chromatin rearrangement, and signal transduction. Notably, regulator proteins play a central role in the process of LLPS. They are essential for the formation, stabilization, and maintenance of the dynamic properties of LLPS, ensuring an appropriate phase separation response to cellular signals. Targeting these regulator proteins is the key to manipulating LLPS for applications in biotechnology, materials science, and medicine, including biomaterials, drug delivery, diagnostics, and synthetic biology. Given their importance, this study focused on an artificial intelligence-based approach to identify regulator proteins in LLPS. We constructed a dataset of 913 positive and 6584 negative protein sequences, and divided it into eight balanced training datasets and a test dataset. Semantic information from protein sequences was extracted using the ESM2_t36 pretrained protein language model, followed by training a multilayer perceptron classifier. The model achieved 0.78 accuracy on the test dataset, outperforming traditional sequence-based methods, one-hot encoding, and other pretrained embedding methods. SHapley Additive exPlanations (SHAP)-based interpretation revealed key biophysical patterns enriched in regulator proteins, including higher levels of charged and disordered residues. Our results show that deep contextual protein representations combined with neural network-based classifiers can accurately identify LLPS regulator proteins. This tool offers new opportunities for understanding condensate biology and designing synthetic phase-separating systems. All data and code are available at: https://github.com/bioplusAI/LLPS_regulators_pred.

Protein–protein interactions (PPIs) perform significant functions in many biological activities likewise gene regulation, metabolic pathways and signal transduction. The deregulation of PPIs may cause deadly diseases, such as cancer, autoimmune, pernicious anaemia etc. Detecting PPIs can aid in elucidating the cellular process's underlying molecular mechanisms and contribute to facilitating the discovery of new proteins for the development of novel drugs. Although high‐throughput wet‐lab technologies have been matured to identify large scale PPI identification; however, the traditional experimental methods are costly and slow and resource intensive. To support experimental techniques, numerous computational approaches have been emerged for identifying PPIs solely from protein sequences. However, the performance of available PPI tools are unsatisfactory and gaps remain for further improvement. In this study, a novel deep learning‐based model, Deep_PPI, was developed for predicting multiple species PPIs. To extract the biological features, the authors used 21D vector representing 20 kinds' native and one special amino acid residue and implemented the Keras binary profile encoding technique to formulate each residue in proteins. The binary profile use the PaddVal strategy to equalise the length of positive and negative PPIs. After extracting the features, the authors fed them into one dimension convolutional neural network to build the final prediction model. The proposed Deep_PPI model, which consider the protein pairs into two convolutional heads. Finally, the authors concatenated the two outputs were concatenated from two branches concatenated by fully connected layer. The efficiency of the proposed predictor was demonstrated both on the cross validation and tested on various species datasets, for example, that is (Human, C. elegans , E. coli , and H. sapiens ). The proposed model surpassed both the machine‐learning models and existing state‐of‐the‐art PPI methods. The proposed Deep_PPI will serve as valuable tool in the discovery of large‐scale PPIs in particular and provide insights for drugs development in general.

Hypertension, often known as high blood pressure, is a major concern to millions of individuals globally. Recent studies have demonstrated the significant efficacy of naturally derived peptides in reducing blood pressure. Hypertension is one of the risks associated with cardiovascular disorders and other health problems. Naturally sourced bioactive peptides possessing antihypertensive properties provide considerable potential as viable substitutes for conventional pharmaceutical medications. Currently, thorough examination of antihypertensive peptide (AHTPs), by using traditional wet‐lab methods is highly expensive and labours. Therefore, in‐silico approaches especially machine‐learning (ML) algorithms are favourable due to saving time and cost in the discovery of AHTPs. In this study, a novel ML‐based predictor, called StackAHTP was developed for predicting accurate AHTPs from sequence only. The proposed method, utilise two types of feature descriptors Pseudo‐Amino Acid Composition and Dipeptide Composition to encode the local and global hidden information from peptide sequences. Furthermore, the encoded features are serially merged and ranked through SHapley Additive explanations (SHAP) algorithm. Then, the top ranked are fed into three different ensemble classifiers (Bagging, Boosting, and Stacking) for enhancing the prediction performance of the model. The StackAHTPs method achieved superior performance compare to other ML classifiers (AdaBoost, XGBoost and Light Gradient Boosting (LightGBM), Bagging and Boosting) on 10‐fold cross validation and independent test. The experimental outcomes demonstrate that our proposed method outperformed the existing methods and achieved an accuracy of 92.25% and F1‐score of 89.67% on independent test for predicting AHTPs and non‐AHTPs. The authors believe this research will remarkably contribute in predicting large‐scale characterisation of AHTPs and accelerate the drug discovery process. At https://github.com/ali‐ghulam/StackAHTPs you may find datasets features used.

The prevalence of lung cancer is increasing due to different risk factors with lower survival rate. Lung cancer complexity exhibits in its metastasizing ability to brain, bones, and adrenal glands. Therefore, despite the advancement of the field, the entire extent of the genetic variations associated with lung metastasis is still elusive. Patients with lung cancer metastasis show decreased effectiveness or resistance towards the available conventional drugs. In the present article, we used Genomic Data Commons (GDC) to explore the primary lung cancer genes with higher tendency to cause metastasis and aim to investigate the compatible drugs. Based on RNA-seq analysis, TFF2 and REG4 were the top two upregulated differentially expressed genes that we investigated in more details for its affinity with Gefitinib - a first generation FDA approved tyrosine kinase inhibitor drug on these two genes. Molecular docking analysis revealed that these two genes show lower binding affinity to Gefitinib, making it less effective for patients with mutation in these genes compared to patients without mutation. We believe our findings would provide better understanding of the primary mutated genes that could lead to lung cancer metastasizing to other organs and effective drug options in the clinical setting.