Media coverage

Coverage of a method that mines causal relationships between diseases from the literature and uses them to improve polygenic risk scores.

KAUST coverage of JaSaPaGe, a population-specific pangenome reference for Saudi and Japanese populations built to improve variant detection and diagnosis for groups underrepresented in existing genome references.

Press coverage of the KAUST/Tufts/JIHS pangenome reference (JaSaPaGe) for Saudi Arabian and Japanese populations, improving variant calling to help close the diagnostic gap for underrepresented populations.

Feature on KSA001, the first complete reference-quality genome from a Saudi individual, and its expansion into a public Saudi pangenome (the group's bio-ontology-research-group/KSA001 resource), addressing underrepresentation of Arab populations in genomics.

KAUST press release on mining causal disease-disease relationships from the literature to improve polygenic risk prediction.

Medical Xpress coverage of causal disease-disease relationship mining for improved polygenic risk scores (syndication of the KAUST/EurekAlert release).

StudyFinds feature on the group's causal map linking thousands of diseases and its use for polygenic risk prediction.

Coverage of DeepGO-SE, which combines protein language models with logical inference over the Gene Ontology to predict protein function, ranking among the top methods in the international CAFA assessment.

KAUST press release on DeepGO-SE, combining protein language models with logical inference over the Gene Ontology for protein function prediction.

Phys.org coverage of DeepGO-SE protein function prediction (syndication of the KAUST/EurekAlert release).

SciTechDaily coverage of DeepGO-SE protein function prediction (syndication of the KAUST/EurekAlert release).

Neuroscience News coverage of DeepGO-SE protein function prediction (syndication of the KAUST/EurekAlert release).

Coverage of STARVar, a tool that ranks disease-associated genetic variants from patient symptoms recorded in standardized or natural-language form, outperforming other variant-prioritization tools.

KAUST press release on STARVar, ranking disease-associated variants from free-text patient symptoms.

Medical Xpress coverage of STARVar, ranking disease variants from free-text patient symptoms (syndication of the KAUST/EurekAlert release).

News-Medical coverage of STARVar for rare-disease variant prioritization (syndication of the KAUST/EurekAlert release).

Coverage of a study quantifying how much model-organism phenotype data improves computational identification of human disease genes.

The Bio-Ontology Research Group team wins the KAUST-NVIDIA GPU Hackathon 2022 for GPU-distributed multimodal protein function prediction.

Coverage of a machine-learning approach that predicts tumor-driving genes from biological features rather than sequence data alone, using tumor sequencing as validation.

BORG alumna Imane Boudellioua receives a KACST national prize for her work on phenotype-based variant prioritization.

KAUST press release on ontology-based machine learning that predicts tumor-driving genes from biological features rather than sequence alone.

Medical Xpress coverage of ontology-based prediction of cancer driver genes (syndication of the KAUST/EurekAlert release).

Coverage of PathoPhenoDB, a public database from the group linking human pathogens to the clinical phenotypes of the diseases they cause, in support of infectious-disease research.

Coverage of the group's embedding methods that translate symbolic, human-readable ontology knowledge into vector spaces so neural networks can combine logical reasoning with statistical learning.

Tech Xplore coverage of the group's ontology embeddings that bridge symbolic AI and machine learning (syndication of the KAUST release).

Medical Xpress coverage of PathoPhenoDB, linking pathogens to disease phenotypes (syndication of the KAUST release).

Coverage of the PhenomeNET Variant Predictor (PVP), which prioritizes disease-causing genomic variants by reasoning over phenotype ontologies and model-organism data.

Coverage of phenotype-driven variant prioritization for personalized medicine, ranking candidate disease variants for individual patients.

Coverage of text-mining work that extracts disease phenotypes from the biomedical literature to build a computational map of human diseases (the human diseasome).