This work introduces MolBioKG, a two-layer system that grounds unseen molecules in biomedical evidence via multi-resolution structural anchoring and outperforms strong baselines across in-graph link recovery, complex multi-hop reasoning, and out-of-graph generalization.
Abstract
Biomedical knowledge graphs (KGs) accelerate drug discovery, but standard pipelines assume query molecules already exist as graph entities, leaving unregistered molecules disconnected. We address this cold-start challenge, termed the out-of-graph molecule problem, by introducing MolBioKG. This two-layer system grounds unseen molecules in biomedical evidence via multi-resolution structural anchoring. It connects an index of 2.74 million molecules (represented by scaffolds, fragments, functional groups, and fingerprints) to a 9.6-million-edge KG. Given only a SMILES string, MolBioKG retrieves structurally related graph entities and traverses their biomedical neighborhoods without task-specific training. It features two inference mechanisms: static multi-anchor retrieval using Reciprocal Rank Fusion, and Adapt-KG, a tool-using LLM policy for adaptive traversal. Evaluated across in-graph link recovery, complex multi-hop reasoning, and out-of-graph generalization, MolBioKG outperforms strong baselines. Notably, it raises Hits@10 from 0.585 to 0.876 in multi-hop reasoning and out-of-graph target recall from 0.145 to 0.269, all while ensuring predictions retain traceable structural anchors and source-attributed KG evidence.
Accurately predicting disease–protein relations in biomedical knowledge graphs helps link disease phenotypes to molecular mechanisms and supports disease-related knowledge discovery and candidate target identification. Biomedical knowledge graphs organize multisource biomedical knowledge, including diseases, proteins, drugs, and pathways, as entity nodes and relational edges, providing a structured foundation for modeling complex biomedical associations. Existing methods are often constrained by single-modality modeling, shallow graph–text fusion, and insufficient traceable evidence, which limits their ability to exploit graph–text complementarity and weakens downstream validation and structural evidence interpretation.
To address these limitations, we propose NEOGRAN, a graph–text collaborative framework comprising three core modules for relation prediction in biomedical knowledge graphs. The dual-encoder architecture captures graph structural patterns and biomedical entity representations to mitigate single-modality modeling. The bidirectional cross-attention module enables deep graph–text interaction to overcome shallow fusion. The interpretable path module generates traceable evidence paths to support prediction verification, structural evidence interpretation, and hypothesis generation. On PrimeKG, NEOGRAN achieved an AUPR of 0.9860 and an AUROC of 0.9875 under the 1:1 sampled classification setting, and further obtained an MRR of 0.0735 in the all-candidate ranking evaluation. External validation on BioKG further shows that NEOGRAN remains effective under differences in entity coverage, relation composition, and local topology, supporting its method-level generalizability across knowledge graph sources.
NEOGRAN provides an effective solution for relation prediction in biomedical knowledge graphs while offering traceable structural evidence for hypothesis generation and further biological validation.
VitaGraph is presented, a comprehensive multi-purpose biological knowledge graph built by integrating and refining multiple public datasets and enabling benchmarking of graph-based models and offering the opportunity to tackle tasks such as drug repurposing, PPI prediction, and side-effect prediction, among others.
Francesco Madeddu, Lucia Testa, Gianluca De Carlo et al.· Scientific Data· 0 citations
This is the first method to expose GNN-derived attributions to an LLM as evidence for property prediction, and achieves the best overall results among generalist models and narrows the gap to specialist models tuned for each task.
Junwoo Park, Minyoung Shin, C. Lee et al.· 0 citations
Initial experiments on heart-failure-focused clinical question answering show that CGX improves evidence retrieval quality and perceived answer reliability over conventional retrieval methods, while reducing total graph construction time by 69.7% under the same input corpus and hardware setting.
Dat Nguyen, Anh N Le, Binh T. D. Trinh et al.· Journal of Biomedical Inform...· 0 citations
Existing general-purpose biomedical knowledge graphs tend to focus on disease mechanisms and drug repurposing, leaving multiomic and wellness-relevant content underrepresented. KRAKEN (Knowledge Research & Analysis Kit for Evidence Networks) addresses this gap by integrating existing graphs (including Translator KG Open, RTX-KG2, and ROBOKOP) with specialized sources such as RefMet, LIPID MAPS, NIH Common Data Elements, Polygenic Score Catalog, and derived wellness measures including biological age and biological BMI. The resulting graph spans ∼15M nodes and ∼113M edges across 62 entity types. KRAKEN adopts the Biolink Model as its semantic layer, ensuring compatibility with standardized resources emerging from the NIH NCATS Biomedical Data Translator program. A lightweight, modular build system rebuilds the full graph (including entity resolution), with peak memory consumption <48 GB, and supports flexible inclusion or exclusion of sources, allowing the user to scope the graph to a domain of interest. Built-in analytical tools include multi-hop reasoning, subgraph extraction, text, vector and hybrid entity search, and enrichment analyses, all accessible through an interactive web interface, a REST API, and a Model Context Protocol server, the last enabling direct consumption by agentic and LLM-based systems. KRAKEN is freely available at https://app.krakenkg.com. GRAPHICAL ABSTRACT
Amy K. Glen, D. Witherington, Trent Leslie et al.· bioRxiv· 0 citations
Knowledge graph (KG) construction pipelines must continuously integrate newly arriving entities into a growing graph. Unlike inserting triples between existing nodes, a newly arriving entity has no graph connectivity: it emerges from the acquisition phase as a raw feature vector and must be assigned to a semantic community before entity resolution and link prediction can operate over a tractable candidate set. Existing multi-view graph clustering methods exploit multiple relation types as structural views, but are transductive: they assume a fixed graph and cannot assign unseen entities without retraining. We propose SNAP-KG (Streaming Node Assignment via Projection for Knowledge Graph Entity Integration), a framework supporting graph-structural multi-view relational clustering and inductive inference for streaming entities. SNAP-KG trains a projector to map a new entity directly to the learned embedding space using only raw features, enabling immediate cluster assignment without graph access or model retraining. Experiments on five benchmark multi-view graph datasets and a production-scale KG of 2.4 million nodes demonstrate multiple orders-of-magnitude inference speedups over retraining-based approaches and competitive clustering quality. As a candidate scoping mechanism for downstream tasks, SNAP-KG achieves 62-75% candidate search reduction on the five benchmark datasets and 97% on OGB-WikiKG2 for entity resolution and link prediction.
Jui-Chien Lin, Mohammad Mohammadi Amiri, O. Seneviratne· 0 citations