The simultaneous use of two or more drugs in clinical treatment may raise the risk of a drug–drug interaction (DDI). DDI prediction is very important to avoid adverse drug events in combination therapy. Recently, deep learning methods have been applied successfully to DDI prediction and improved prediction performance. However, there are still some problems with the present models, such as low accuracy due to information loss during molecular representation or incomplete drug feature mining during the training process. Aiming at these problems, this study proposes an accurate and simple framework named PTB-DDI for drug–drug interaction prediction. The PTB-DDI framework consists of four key modules: (1) ChemBerta tokenizer for molecular representation, (2) Bidirectional Long Short-Term Memory (BiLSTM) to capture the bidirectional context-aware features of drugs, (3) Multilayer Perceptron (MLP) for mining the nonlinear relationship of drug features, and (4) interaction predictor to perform an affine transformation and final prediction. In addition, we investigate the effect of dual-mode on parameter-sharing and parameter-independent within the PTB-DDI framework. Furthermore, we conducted comprehensive experiments on the two real-world datasets (i.e., BIOSNAP and DrugBank) to evaluate PTB-DDI framework performance. The results show that our proposed framework has significant improvements over the baselines based on both datasets. Based on the BIOSNAP dataset, the AUC-ROC, PR-AUC, and F1 scores are 0.997, 0.995, and 0.984, respectively. These metrics are 0.896, 0.873, and 0.826 based on the DrugBank dataset. Then, we conduct the case studies on the three newly approved drugs by the Food and Drug Administration (FDA) in 2024 using the PTB-DDI framework in dual modes. The obtained results indicate that our proposed framework has advantages for predicting drug–drug interactions and that the dual modes of the framework complement each other. Furthermore, a free website is developed to enhance accessibility and user experience.
2 April 2025 in Journal of Chemical Information and Modeling
Leukemia is a malignant tumor with a high recurrence rate and poor prognosis for patients. Thus, there is an urgent need to explore new therapeutic targets that play critical roles in leukemogenesis but have little effect on normal hematopoietic cells. Here, we show that RNA binding protein with multiple splicing (RBPMS), which is highly expressed in acute myeloid leukemia (AML) and associated with poor prognosis of AML, plays critical roles in leukemogenesis. Our study shows that inhibition of RBPMS inhibits self-renewal of leukemia-initiating cells (LICs) and leukemia development but has little effect on normal hematopoiesis. Mechanistically, RBPMS recruits the N6 -methyladenosine (m 6 A) reader insulin-like growth factor 2 mRNA binding protein 3 (IGF2BP3), which promotes the stability of the forkhead box O1 ( FOXO1 ) mRNA in an m 6 A-dependent manner. Moreover, RBPMS contributes to the progression of leukemia by directly binding to FOXO1 and promoting FOXO1-regulated glycolysis. Overexpression of FOXO1 has been shown to reverse RBPMS inhibition–induced phenotypes in both leukemic cells and mouse models. We also designed a specific inhibitor of RBPMS that has therapeutic effects in AML patient-derived xenograft (PDX) models. We therefore highlight RBPMS as a promising drug target for leukemia therapy.
Voltage-gated potassium channel KCNQ1 (Kv7.1) underpins cardiac repolarization, epithelial ion transport, and inner ear function. Its versatility arises from interactions with KCNE proteins, calmodulin (CaM), and the lipid phosphatidylinositol 4,5-bisphosphate (PIP2), yet the molecular basis of PIP2 regulation remains incompletely understood. Here, we present the Stepwise Integrated Multi-scale Dynamics and Advanced Analysis (SIMDA) framework, which integrates coarse-grained and all-atom molecular dynamics, well-tempered metadynamics, and clustering and energy analyses. Over 2,000 µs of simulations across eight functional states find six recurrent PIP2 sites (C0-C5), each complex populating three to four. Sites C1 and C3 agree with experimental densities, whereas C0, C4, and C5 are forward predictions. KCNE3 stabilizes the primary C1 site by amplifying a C-terminal twist motion, while CaM tunes binding at C4 and C5. Interconnected transfer pathways form a dynamic circular route among sites. SIMDA thus provides a generalizable strategy for dissecting lipid-protein dynamics.