Protein phosphorylation is an important type of post-translational modification that regulates various activities of cell life inside human body. The accurate identification of phosphorylation sites can provide new insights for revealing the specific function of protein. However, it is time-consuming and inefficient to apply the experiment-based techniques in investigating the phosphorylation sites in proteins. Additionally, computational approaches are regarded as an ideal choice in such a big data era. Therefore, in this work, we designed a new computational method to identify phosphorylation sites. At first, phosphorylation data was collected from human proteins to construct an objective and strict benchmark dataset. By a series of feature analysis, we found that the distributions of conservation scores and nine physicochemical properties surrounding the phosphorylation sites in positive samples are significantly different from those surrounding non-phosphorylation sites in negative samples. Based on these features, a novel sequence-based method for predicting the phosphorylation sites in human proteomics was proposed, which incorporated the conservation scores with position-associated attributes that reflect the correlation of physicochemical characteristics among amino acid residues. Furthermore, the analysis of variance (ANOVA) was utilized to obtain the optimal feature subset which could produce the highest accuracy. Comparison with the published predictor demonstrated the superiority of our predictor. Finally, a user-friendly online tool called iPhoPred was established and can be freely available at http://lin-group.cn/server/iPhoPred/. We hope the tool will provide important guide for the study of protein phosphorylation.

ScRiskCell is an interpretable intelligent computational framework that leverages nearly 500,000 islet cell expression profiles from 106 donors across different continuous disease states. By calculating the intrinsic relationship between donor disease states and cell expression profiles, it assigns a pseudo‐cell state index to each cell. Sorting the pseudo‐indexes of cells enables the identification of risk cells truly disrupted by the disease. Importantly, scRiskCell reveals the dynamic aggregation pattern of risk cells during disease progression, providing mechanistic insights for early disease prediction and clinical dynamic monitoring of disease progression.

RepliChrom is an interpretable machine learning model that predicts enhancer‐promoter interactions using DNA replication timing across multiple cell types. By integrating replication timing with chromatin interaction data from multiple experimental platforms, it accurately distinguishes true interactions and reveals promoter‐region signals as key regulatory drivers. Importantly, the RepliChrom uncovers cancer‐specific chromatin patterns in leukemia, offering mechanistic insights into how replication timing shapes long‐range gene regulation in both normal and diseased genomes.

Background: Dimension disaster is often associated with feature extraction. The extracted features may contain more redundant feature information, which leads to the limitation of computing ability and overfitting problems. Objective: Feature selection is an important strategy to overcome the problems from dimension disaster. In most machine learning tasks, features determine the upper limit of the model performance. Therefore, more and more feature selection methods should be developed to optimize redundant features. Methods: In this paper, we introduce a new technique to optimize sequence features based on the binomial distribution (BD). Firstly, the principle of the binomial distribution algorithm is introduced in detail. Then, the proposed algorithm is compared with other commonly used feature selection methods on three different types of datasets by using a Random Forest classifier with the same parameters. Results: The results confirm that BD has a promising improvement in feature selection and classification accuracy. Conclusion: Finally, we provide the source code and executable program package (http: //lingroup. cn/server/BDselect/), by which users can easily perform our algorithm in their researches.

Lactylation is a newly discovered type of post-translational modification, primarily occurring on lysine (K) residues of both histones and non-histones to exert diverse effects on target proteins. Research has shown that lysine lactylation (Kla) modification is ubiquitous in different cells and participates in the determination of cell function and fate, as well as in the initiation and progression of various diseases. Precise identification of Kla sites is fundamental for elucidating their biological functions and uncovering their application potential. Here, we proposed a novel human Kla site predictor (named PBertKla) through curating a reliable benchmark dataset with proper sample length and sequence identity threshold to train a protein large language model with optimal hyperparameters. Extensive experimental results consistently demonstrated that our model possessed robust human Kla site prediction ability, achieving an AUC (area under receiver operating characteristic curve) value of over 0.880 on the independent validation data. Feature visualization analysis further validated the effectiveness of in feature learning and representation from Kla sequences. Moreover, we benchmarked PBertKla against other cutting-edge models on an independent testing dataset from different sources, highlighting its superiority and transferability. All results indicated that PBertKla excelled as an automatic predictor of human Kla sites, and it would advance the investigation of lactylation modifications and their significance in health and disease.

Type 2 diabetes mellitus (T2DM) is a prevalent and complex metabolic disorder characterized by insulin resistance, progressive β-cell dysfunction, and severe systemic complications. Advances in single-cell multi-omics—transcriptomics, chromatin accessibility profiling, and integrative analyses—have offered unprecedented insights into the cellular heterogeneity and regulatory networks of pancreatic islets. We highlight recent discoveries in islet cell heterogeneity and β-cell pathophysiology, with a particular focus on dysfunction and dedifferentiation. We further underscore the computational frameworks that enable these discoveries, spanning data preprocessing, multi-omics integration, and machine learning-driven analyses, which collectively enable the dissection of disease-relevant cell subpopulations and the reconstruction of developmental and regulatory trajectories. We also examine how impaired signaling within islets and chronic adipose inflammation contribute to T2DM pathogenesis. Finally, we discuss key challenges in clinical translation—including limited population diversity in single-cell atlases and the interpretability of computational models—and propose future directions toward precision diagnostics and therapeutic innovation in T2DM.