Research
BORG works on biomedical ontologies, neuro-symbolic AI, disease and phenotype informatics, and protein-function prediction. The tabs below cover the research topics we work on, the funded projects that support that work, and the open-source software the group produces.
Research topics the Bio-Ontology Research Group works on. Open a topic for the full overview, related projects, software, and publications.
Neuro-symbolic AI
I work on methods that integrate symbolic knowledge with statistical learning. This includes mapping entities in formal ontologies into vector spaces while preserving their semantic relations. I develop embedding frameworks for Description Logics (e.g., EL++ and ALC) that provide mathematical guarantees for logical soundness and approximate the interpretation of formalized theories.
Ontology engineering
I develop architectures for processing massive, heterogeneous data using Semantic Web standards. This work includes the AberOWL infrastructure for ontology-based data access and methods for establishing interoperability across distributed databases through linked knowledge graphs.
Applied Ontology
I use ontologies to standardize and analyze complex phenotypes across domains. My earlier work focused on foundational ontologies, including the General Formal Ontology (GFO) and its biological extension GFO-Bio, as well as the development of a formal ontology of functions to curate functional knowledge in the life sciences.
Protein function
Large-scale ontologies like the Gene Ontology (GO) provide essential background knowledge for understanding protein activity. I work on the DeepGO family of systems, which utilize formalized axioms to constrain deep learning models for protein function prediction. These systems are used to derive functional insights from sequence and interaction data.
Rare disease
The diagnosis of rare diseases requires the integration of patient-specific data with large-scale background knowledge, such as the Human Phenotype Ontology (HPO). I develop systems like PhenomeNET and PVP that use automated reasoning and machine learning to prioritize disease-causing genomic variants based on their phenotypic consequences.
Drug mechanisms
I apply ontologies and knowledge graphs to model drug-target interactions, drug indications, and adverse drug reactions. This work links molecular data to systems biology through causal knowledge graphs, enabling the identification of mechanistic relationships and potential drug repurposing targets.
Genomics
I contribute to the development of genomic resources and the analysis of population-specific genomic data. This includes reference genome assemblies, pangenome graphs for the Saudi and wider Middle Eastern population, variant-calling and structural-variant pipelines, and the analysis of antimicrobial resistance from whole-genome sequencing.
Biomedical informatics
I work on biomedical informatics infrastructure that turns research-grade data into usable inputs for clinicians and researchers. This includes biomedical knowledge-base construction (PathoPhenoDB, PhenomeNET, PhenomeBrowser), text-mining of biomedical literature, integration of clinical phenotype encodings, and analytics over electronic health records.
Semantic similarity
I develop and benchmark semantic similarity measures over biomedical ontologies, including measures that operate on the OWL axiomatic structure of an ontology rather than only on its lexical or taxonomic skeleton. These measures underpin phenotype-based disease gene prioritization, ontology-aware protein function transfer, and biodiversity knowledge graph search.
Microbial communities
I develop methods that lift single-protein function prediction up to the level of microbial communities, combining ontology-aware deep learning with multi-scale systems approaches. Applications include desert-soil and mangrove-microbiome design, bioprospecting from Saudi-Arabian extremophile habitats, and metagenomics-driven functional characterisation of patient and environmental microbiomes.
Phenotype informatics
I develop the informatics infrastructure for phenotype data across species and clinical settings: phenotype ontologies (HPO, MP, ZP, FLOPO, plant traits), cross-species phenotype crosswalks, tools that capture and standardise phenotype descriptions, and computational pipelines that link phenotype data to underlying genes, variants, and diseases.
Bioengineering
I contribute computational and omics analysis to collaborative bioengineering projects. Examples include analysing transcriptomic and metabolomic responses of cells cultured in biomimetic peptide scaffolds, patient-derived disease-model analysis for precision medicine, and the integration of multi-omics data with engineered biological systems.
Funded research projects led by or involving the Bio-Ontology Research Group.
Pre-KAUST projects
- (2005–2009) Basic considerations for improving interoperability between ontology-based biological information systems
- (2009–2010) Postdoctoral research on biomedical ontology reasoning and integration
- (2010–2014) Phenotype ontologies and translational research (Cambridge / Aberystwyth)
Open-source tools, libraries, services, and ontologies maintained by the Bio-Ontology Research Group. Source is on github.com/bio-ontology-research-group.
Ontology Embedding & Machine Learning
Libraries and methods that turn ontologies into vector representations or otherwise combine logical structure with statistical learning.
mOWL
Python library for machine learning with biomedical ontologies. Unifies projection-, axiom- and geometric-embedding methods (EL Embeddings, ELBE, BoxSquaredEL, OWL2Vec*, DL2Vec, OPA2Vec) behind one API, with first-class OWLAPI access and PyTorch integration.
Walking RDF and OWL
Original feature-learning method over RDF graphs and OWL ontologies via biased random walks; the seed implementation for many later embedding methods including OWL2Vec*.
OPA2Vec
Combines ontology axioms with associated annotation properties (labels, synonyms, definitions) into a single corpus, then trains Word2Vec to produce semantically rich vectors for ontology classes.
EL Embeddings
Reference implementation of geometric embeddings for the EL++ description logic, the predecessor of GeometrE and BoxSquaredEL. Preserves subsumption reasoning by mapping classes to convex regions.
Interpretable Learning
Generates interpretable symbolic rules from learned representations over biomedical knowledge bases.
Protein Function Prediction
Deep-learning and neuro-symbolic models for predicting Gene Ontology functional annotations of proteins.
DeepGOPlus
CNN-ensemble protein-function predictor that augments sequence-based scoring with k-nearest-neighbour homology and GO axioms; one of the strongest CAFA-evaluated open models.
DeepGO
Original sequence-based, ontology-aware deep classifier for predicting Gene Ontology functional annotations; basis of the entire DeepGO family of tools.
DeepGOZero
Zero-shot extension of DeepGO using model-theoretic ELEmbeddings to predict GO classes that have never been observed during training.
DeepGOMeta
DeepGO trained specifically for metagenomic communities; predicts functional roles of proteins recovered from environmental samples and links them to biogeochemical processes.
PhenoGoCon
Predicts gene–phenotype associations from predicted Gene Ontology functions; bridges GO function prediction and HPO/MPO phenotype prediction.
Genomic Context
Bacterial protein-function prediction that exploits operon and genome-neighbourhood structure in addition to sequence and homology.
Variant and Disease Prioritization
Tools that combine phenotype, ontology, and sequence data to rank candidate disease-causing variants and predict gene-disease associations.
PhenomeNET-VP
Phenotype-driven variant prioritization for whole-exome and whole-genome sequencing data; widely used implementation of the phenotype-aware variant ranking approach.
DeepSVP
Prioritizes structural and copy-number variants by combining patient phenotype with gene-function similarity learned from biomedical ontologies.
GenomeLinter
AI-powered clinical decision-support tool that ingests annotated VCFs and synthesises diagnostic interpretations for rare-disease patients without requiring deep bioinformatics expertise.
Ontology Reasoning & Tooling
Reasoning-as-a-service infrastructure, ontology repositories, and utilities for working with OWL.
vec2SPARQL
Adds embedding-similarity functions to a SPARQL endpoint so that vector-space queries (k-nearest neighbours, cosine similarity) can be mixed with classical graph patterns.
Onto2Graph
Generates entailment-aware graph projections of OWL ontologies suitable for downstream graph machine learning while preserving the axioms' deductive structure.
AberOWL
Ontology repository delivering OWL EL reasoning as a service: stores hundreds of bio-ontologies, exposes SPARQL with class-expression query expansion, and powers semantic search over PubMed/PMC.
OntoFunc
Ontology-driven enrichment analysis that supports arbitrary OWL ontologies and full subsumption-aware aggregation, not only GO.
Knowledge Graphs & Drug Discovery
Biomedical knowledge graph construction, drug repurposing, drug-drug interactions, and graph-based prediction.
Multi-Drug Embedding
Drug repurposing method that learns joint embeddings of drugs, targets and diseases from biomedical knowledge graphs and the scientific literature.
NanoDesigner
Iterative refinement framework for nanobody/CDR design that explicitly models the antigen–CDR interdependence; companion code to the NanoDesigner paper.
SmuDGE
Semantic disease-gene embeddings; integrates phenotype, function and pathway ontologies into a unified vector space for downstream prediction.
PathoPhenoDB
Curated database of pathogens and the disease phenotypes they cause, distributed as an OWL ontology and an interactive web application.
Teaching & Tutorials
Self-contained teaching material accompanying our courses and review articles.
Machine Learning with Ontologies
Companion code and worked examples for the Briefings in Bioinformatics tutorial review; the most-starred repository in the group.
Ontology Tutorial
Hands-on tutorial that walks new users through OWL, automated reasoning, and ontology-aware data analysis; basis for the AI in Biomedicine summer school.
mOWL Tutorial
Step-by-step worked notebooks that demonstrate every embedding family in mOWL on protein-function, gene-disease, and ontology-completion tasks.
Ontologies & Resources
Bio-ontologies and curated resources maintained by the group.
Units of Measurement Ontology (UO)
OBO Foundry ontology of units of measurement; aligned with QUDT and used across biomedical data standards.
PhenomeNet
Cross-species phenotype ontology and similarity network combining HPO, MPO, ZP and others; the substrate behind PhenomeNET-VP and DeepPheno.