CAREPath: New AI Framework Improves Drug Repurposing by Balancing Mechanistic Reasoning and Context
Researchers have developed CAREPath, a computational framework that integrates biomedical knowledge graphs with large language models to identify new uses for existing drugs. The system balances two search strategies—one focused on short, mechanistically specific drug-gene-disease paths and another that captures broader biological context from gene neighborhoods—outperforming 18 baseline methods across five biomedical knowledge graphs. The approach could accelerate drug repurposing by providing interpretable, mechanism-aware predictions that align with known biological functions.
CAREPath is a new knowledge graph and language model (KG-LLM) framework designed to address key limitations in computational drug repurposing. Existing methods that traverse biomedical knowledge graphs face a trade-off: longer paths grow combinatorially and often pass through highly connected hub genes, introducing irrelevant signals, while shorter or sparser paths may miss important biological context. CAREPath resolves this by combining a depth-first search (DFS)-like module—which restricts traversal to short disease-gene-drug paths and encodes them as semantic embeddings using a biomedical language model—with a breadth-first search (BFS)-like module that builds broader mechanism-context embeddings from one-hop gene neighborhoods, augmented by pharmacologically related drugs and gene-signature-similar diseases. Evaluated across five biomedical knowledge graphs against 18 baseline methods, CAREPath achieved the best overall area under the precision-recall curve (AUPRC), with improvements of up to 3.8%. Analysis showed that semantic short-path encoding was the largest contributor to performance, while the mechanism-context augmentation module improved robustness when evidence was sparse and strengthened alignment with Gene Ontology functional annotations. Case studies involving recently FDA-approved drug indications further supported the framework's practical relevance, and the source code has been made publicly available.
What's missing
As a preprint, this work has not yet undergone formal peer review. The study does not report external prospective validation—i.e., whether CAREPath's novel repurposing predictions have been experimentally tested in vitro or in vivo. The degree to which AUPRC improvements of up to 3.8% translate into meaningfully better real-world drug candidate prioritization remains an open question. Additionally, computational cost and scalability to very large knowledge graphs beyond the five tested are not fully characterized.
What different sources said
- bioRxivCenter
CAREPath: Semantic Context-Aware Reasoning Paths with Mechanism-Augmented Embeddings for Drug Repurposing
Related
Multiscale Brain Model Predicts Novel Propofol Anesthesia Biomarker Without Training on Clinical Data
Researchers developed a mechanistic computational model of thalamocortical brain circuits that successfully predicted a previously unnoticed dose-dependent biomarker of propofol anesthesia. The model, driven solely by GABA-A receptor modulation, reproduced empirical data from both macaques and humans without being fitted to any anesthesia-specific data. The findings suggest that simulation-first approaches could accelerate biomarker discovery in neuropharmacology without requiring large clinical datasets.
Green-Synthesized Zinc Oxide Nanoparticles from Mimosa pudica Show Biocompatibility with Bone Marrow Stem Cells in Lab Study
Researchers synthesized zinc oxide nanoparticles using Mimosa pudica leaf extract and tested their effects on human bone marrow mesenchymal stromal cells, finding the nanoparticles preserved cell viability, structure, and bone-forming capacity. The plant-derived nanoparticles outperformed both the raw plant extract and conventionally synthesized zinc oxide in maintaining cell metabolic activity over five days. The findings suggest these bioactive nanomaterials could be candidates for musculoskeletal tissue engineering, though the research remains at an early in vitro stage.
Study Compares Genetic Modeling Approaches for Dyadic Social Interactions in Animals
A new preprint study compared two statistical modeling approaches for analyzing the genetic basis of social interactions in animals, finding that dyadic models outperform marginal models that aggregate individual-level data. The research used pig aggression data from 797 finishing pigs across 59 social groups as a test case. The findings have implications for how animal geneticists model and interpret the heritable components of social behavior.