Neuro-Symbolic AI

INDIGENA: inductive prediction of disease–gene associations using phenotype ontologies

Year: 2026 Venue: Bioinformatics Authors: Fernando Zhapa-Camacho, Robert Hoehndorf DOI: 10.1093/bioinformatics/btag325 Abstract MOTIVATION: Predicting gene-disease associations (GDAs) can be framed as a ranking problem where genes are ranked for a query disease based on features such as phenotypic similarity. By describing phenotypes using phenotype ontologies, ontology-based semantic similarity measures can be used. However, traditional semantic similarity measures use only the ontology taxonomy. Recent methods based on ontology embeddings compare phenotypes in latent space; these methods can

Neuro-Symbolic AI in Life Sciences

Year: 2025 Venue: Handbook on Neurosymbolic AI and Knowledge Graphs Authors: Robert Hoehndorf, Catia Pesquita, Fernando Zhapa-Camacho DOI: 10.3233/faia250239 Abstract Life sciences have a long history of driving advancements in various disciplines, including mathematics, philosophy, and logic. In recent years, life sciences have also become a significant application area for Artificial Intelligence (AI) technologies, including for neuro-symbolic AI methods. The life sciences knowledge infrastructure, characterized by its widespread use of ontologies, complex annotation models, large size, and

Fully Geometric Multi-hop Reasoning on Knowledge Graphs with Transitive Relations

Year: 2026 Venue: The Semantic Web -- ESWC 2026 Authors: Fernando Zhapa-Camacho, Robert Hoehndorf DOI: 10.1007/978-3-032-25156-5_14 Abstract Multi-hop logical reasoning on knowledge graphs requires faithfully mapping the logical semantics to latent space. Current geometric embedding methods show to be useful on this task by mapping entities to geometric regions and logical operations to latent transformations. While a geometric embedding can provide a direct interpretability framework for query answering, current methods have only leveraged the geometric construction of entities, failing to

Lattice-Preserving ALC Ontology Embeddings

Year: 2024 Venue: Neural-Symbolic Learning and Reasoning Authors: Fernando Zhapa-Camacho, Robert Hoehndorf DOI: 10.1007/978-3-031-71167-1_19 Abstract Generating vector representations (embeddings) of OWL ontologies is a growing task due to its applications in predicting missing facts and knowledge-enhanced learning in fields such as bioinformatics. The underlying semantics of OWL ontologies is expressed using Description Logics (DLs). Initial approaches to generate embeddings relied on constructing a graph out of ontologies, neglecting the semantics of the logic therein. Recent semantic

Vec2SPARQL: integrating SPARQL queries and knowledge graph embeddings

Year: 2018 Venue: Proceedings of the 11th International Conference Semantic Web Applications and Tools for Life Sciences, SWAT4LS 2018, Antwerp, Belgium, December 3-6, 2018. Authors: Maxat Kulmanov, Senay Kafkas, Andreas Karwath, Alexander Malic, Georgios V. Gkoutos, Michel Dumontier, Robert Hoehndorf Topics Neuro-symbolic AI · Ontology engineering Acknowledged projects ccf-microbial-cell-factories crg-bio2vec

Neural Multi-hop Logical Query Answering with Concept-Level Answers

Year: 2023 Venue: The Semantic Web – ISWC 2023 Authors: Zhenwei Tang, Shichao Pei, Xi Peng, Fuzhen Zhuang, Xiangliang Zhang, Robert Hoehndorf Abstract Neural multi-hop logical query answering ({LQA}) is a fundamental task to explore relational data such as knowledge graphs, which aims at answering multi-hop queries with logical operations based on distributed representations of queries and answers. Although previous {LQA} methods can give specific instance-level answers, they are not able to provide descriptive concept-level answers, where each concept is a description of a set of instances

Positive-Unlabeled Learning with Adversarial Data Augmentation for Knowledge Graph Completion

Year: 2022 Venue: Proceedings of the Thirty-First International Joint Conference on Artificial Intelligence Authors: Zhenwei Tang, Shichao Pei, Zhao Zhang, Yongchun Zhu, Fuzhen Zhuang, Robert Hoehndorf, Xiangliang Zhang DOI: 10.24963/ijcai.2022/312 Abstract Most real-world knowledge graphs (KG) are far from complete and comprehensive. This problem has motivated efforts in predicting the most plausible missing facts to complete a given KG, i.e., knowledge graph completion (KGC). However, existing KGC methods suffer from two main issues, 1) the false negative issue, i.e., the sampled negative

Robust Knowledge Graph Embedding via Denoising

Year: 2026 Venue: The Semantic Web -- ESWC 2026 Authors: Tengwei Song, Xudong Ma, Yang Liu, Jie Luo, Robert Hoehndorf DOI: 10.1007/978-3-032-25156-5_22 Abstract Knowledge graph embedding models have achieved remarkable success in link prediction and reasoning tasks, yet they remain highly vulnerable to perturbations in the embedding space. Such perturbations, whether introduced by noisy triples, representation drift or adversarial manipulation, can lead to severe degradation in prediction stability and significantly affect downstream multi-hop reasoning processes. To address this challenge, we

Ontology-Based Concept Recognition by Using Word Embeddings

Year: 2018 Venue: Bio-Ontologies COSI Authors: Sara Althubaiti, Senay Kafkas, Robert Hoehndorf Topics Neuro-symbolic AI · Biomedical informatics

Semi-Supervised Entity Alignment via Knowledge Graph Embedding with Awareness of Degree Difference

Year: 2019 Venue: The World Wide Web Conference Authors: Shichao Pei, Lu Yu, Robert Hoehndorf, Xiangliang Zhang DOI: 10.1145/3308558.3313646 Abstract Entity alignment associates entities in different knowledge graphs if they are semantically same, and has been successfully used in the knowledge graph construction and connection. Most of the recent solutions for entity alignment are based on knowledge graph embedding, which maps knowledge entities in a low-dimension space where entities are connected with the guidance of prior aligned entity pairs. The study in this paper focuses on two

Enhancing Geometric Ontology Embeddings for ^++ with Negative Sampling and Deductive Closure Filtering

Year: 2024 Venue: Neural-Symbolic Learning and Reasoning Authors: Olga Mashkova, Fernando Zhapa-Camacho, Robert Hoehndorf DOI: 10.1007/978-3-031-71167-1_18 Abstract Ontology embeddings map classes, relations, and individuals in ontologies into \(\mathbb {R}^n\) , and within \(\mathbb {R}^n\) similarity between entities can be computed or new axioms inferred. For ontologies in the Description Logic \(\mathcal{E}\mathcal{L}^{++}\) , several embedding methods have been developed that explicitly generate models of an ontology. However, these methods suffer from some limitations; they do not

EL Embeddings: Geometric construction of models for the Description Logic EL++

Year: 2019 Venue: Proceedings of IJCAI 2019 Authors: Maxat Kulmanov, Wang Liu-Wei, Yuan Yan, Robert Hoehndorf Abstract An embedding is a function that maps entities from one algebraic structure into another while preserving certain characteristics. Embeddings are being used successfully for mapping relational data or text into vector spaces where they can be used for machine learning, similarity search, or similar tasks. We address the problem of finding vector space embeddings for theories in the Description Logic EL++ that are also models of the TBox. To find such embeddings, we define an

Large-Scale Knowledge Integration for Enhanced Molecular Property Prediction

Year: 2024 Venue: Neural-Symbolic Learning and Reasoning Authors: Yasir Ghunaim, Robert Hoehndorf DOI: 10.1007/978-3-031-71170-1_10 Abstract Pre-training machine learning models on molecular properties has proven effective for generating robust and generalizable representations, which is critical for advancements in drug discovery and materials science. While recent work has primarily focused on data-driven approaches, the KANO model introduces a novel paradigm by incorporating knowledge-enhanced pre-training. In this work, we expand upon KANO by integrating the large-scale ChEBI knowledge

From Axioms over Graphs to Vectors, and Back Again: Evaluating the Properties of Graph-based Ontology Embeddings

Year: 2023 Venue: Proceedings of the 17th International Workshop on Neural-Symbolic Learning and Reasoning, La Certosa di Pontignano, Siena, Italy, July 3-5, 2023 Authors: Fernando Zhapa-Camacho, Robert Hoehndorf Abstract Several approaches have been developed that generate embeddings for Description Logic ontologies and use these embeddings in machine learning. One approach of generating ontologies embeddings is by first embedding the ontologies into a graph structure, i.e., introducing a set of nodes and edges for named entities and logical axioms, and then applying a graph embedding to

SIDEKICK: A Semantically Integrated Resource for Drug Effects, Indications, and Contraindications

Year: 2026 Venue: The Semantic Web -- ESWC 2026 Authors: Mohammad Ashhad, Olga Mashkova, Ricardo Henao, Robert Hoehndorf DOI: 10.1007/978-3-032-25159-6_14 Abstract Pharmacovigilance and clinical decision support systems utilize structured drug safety data to guide medical practice. However, existing datasets frequently depend on terminologies such as MedDRA, which limits their semantic reasoning capabilities and their interoperability with Semantic Web ontologies and knowledge graphs. To address this gap, we developed SIDEKICK, a knowledge graph that standardizes drug indications

mOWL: Python library for machine learning with biomedical ontologies

Year: 2023 Venue: Bioinformatics Authors: Fernando Zhapa-Camacho, Maxat Kulmanov, Robert Hoehndorf DOI: 10.1093/bioinformatics/btac811 Abstract Supplementary data are available at Bioinformatics online. Topics Neuro-symbolic AI

Semantic Disease Gene Embeddings (SmuDGE): phenotype-based disease gene prioritization without phenotypes

Year: 2018 Venue: Bioinformatics Authors: Mona Alshahrani, Robert Hoehndorf Abstract Motivation: In the past years, several methods have been developed to incorporate information about phenotypes into computational disease gene prioritization methods. These methods commonly compute the similarity between a disease's (or patient's) phenotypes and a database of gene-to-phenotype associations to find the phenotypically most similar match. A key limitation of these methods is their reliance on knowledge about phenotypes associated with particular genes which is highly incomplete in humans as well

OPA2Vec: combining formal and informal content of biomedical ontologies to improve similarity-based prediction

Year: 2018 Venue: Bioinformatics Authors: Fatima Zohra Smaili, Xin Gao, Robert Hoehndorf DOI: 10.1093/bioinformatics/bty933 Abstract Motivation: Ontologies are widely used in biology for data annotation, integration and analysis. In addition to formally structured axioms, ontologies contain meta-data in the form of annotation axioms which provide valuable pieces of information that characterize ontology classes. Annotation axioms commonly used in ontologies include class labels, descriptions or synonyms. Despite being a rich source of semantic information, the ontology meta-data are generally

Onto2Vec: joint vector-based representation of biological entities and their ontology-based annotations

Year: 2018 Venue: Bioinformatics Authors: Fatima Zohra Smaili, Xin Gao, Robert Hoehndorf Abstract Motivation: Biological knowledge is widely represented in the form of ontology-based annotations: ontologies describe the phenomena assumed to exist within a domain, and the annotations associate a (kind of) biological entity with a set of phenomena within the domain. The structure and information contained in ontologies and their annotations make them valuable for developing machine learning, data analysis and knowledge extraction algorithms; notably, semantic similarity is widely used to

DELE: Deductive EL++ Embeddings for Knowledge Base Completion

Year: 2026 Venue: Neurosymbolic Artificial Intelligence Authors: Olga Mashkova, Fernando Zhapa-Camacho, Robert Hoehndorf DOI: 10.1177/29498732261420011 Abstract Ontology embeddings map classes, roles, and individuals in ontologies into R^n , and within R^n similarity between entities can be computed or new axioms inferred. For ontologies in the Description Logic EL^++ , several optimization-based embedding methods have been developed that explicitly generate models of an ontology. However, these methods suffer from some limitations; they do not distinguish between statements that are

DeepGOZero: improving protein function prediction from sequence and zero-shot learning based on ontology axioms

Year: 2022 Venue: Bioinformatics Authors: Maxat Kulmanov, Robert Hoehndorf DOI: 10.1093/bioinformatics/btac256 Abstract Supplementary data are available at Bioinformatics online. Topics Protein function · Neuro-symbolic AI

DTI-Voodoo: machine learning over interaction networks and ontology-based background knowledge predicts drug interactions

Year: 2021 Venue: Bioinformatics Authors: Tilman Hinnerichs, Robert Hoehndorf DOI: 10.1093/bioinformatics/btab548 Abstract Supplementary data are available at Bioinformatics online. Topics Drug mechanisms · Neuro-symbolic AI

DeepPVP: phenotype-based prioritization of causative variants using deep learning

Year: 2019 Venue: BMC Bioinformatics Authors: Imane Boudellioua, Maxat Kulmanov, Paul N Schofield, Georgios V Gkoutos, Robert Hoehndorf Abstract Background: Prioritization of variants in personal genomic data is a major challenge. Recently, computational methods that rely on comparing phenotype similarity have shown to be useful to identify causative variants. In these methods, pathogenicity prediction is combined with a semantic similarity measure to prioritize not only variants that are likely to be dysfunctional but those that are likely involved in the pathogenesis of a patient{

DeepGO: predicting protein functions from sequence and interactions using a deep ontology-aware classifier

Year: 2017 Venue: Bioinformatics Authors: Maxat Kulmanov, Mohammed Asif Khan, Robert Hoehndorf Abstract Motivation: A large number of protein sequences are becoming available through the application of novel high-throughput sequencing technologies. Experimental functional characterization of these proteins is time-consuming and expensive, and is often only done rigorously for few selected model organisms. Computational function prediction approaches have been suggested to fill this gap. The functions of proteins are classified using the Gene Ontology (GO), which contains over 40 000 classes

Data science and symbolic AI: Synergies, challenges and opportunities

Year: 2017 Venue: Data Science Authors: Robert Hoehndorf, Nuria Queralt-Rosinach DOI: 10.3233/ds-170004 Abstract Symbolic approaches to artificial intelligence represent things within a domain of knowledge through physical symbols, combine symbols into symbol expressions, and manipulate symbols and symbol expressions through inference processes. While a large part of Data Science relies on statistics and applies statistical approaches to artificial intelligence, there is an increasing potential for successfully applying symbolic approaches as well. Symbolic representations and symbolic

Ontology Embedding: A Survey of Methods, Applications and Resources

Year: 2025 Venue: IEEE Transactions on Knowledge and Data Engineering Authors: Jiaoyan Chen, Olga Mashkova, Fernando Zhapa-Camacho, Robert Hoehndorf, Yuan He, Ian Horrocks DOI: 10.1109/tkde.2025.3559023 Abstract Ontologies are widely used for representing domain knowledge and meta data, playing an increasingly important role in Information Systems, the Semantic Web, Bioinformatics and many other domains. However, logical reasoning that ontologies can directly support are quite limited in learning, approximation and prediction. One straightforward solution is to integrate statistical analysis

Prioritizing genomic variants through neuro-symbolic, knowledge-enhanced learning

Year: 2024 Venue: Bioinformatics Authors: Azza Althagafi, Fernando Zhapa-Camacho, Robert Hoehndorf DOI: 10.1093/bioinformatics/btae301 Abstract EmbedPVP and all evaluation experiments are freely available at https://github.com/bio-ontology-research-group/EmbedPVP. Topics Rare disease · Neuro-symbolic AI

LEP-AD: language embedding of proteins and attention to drugs predicts drug-target interactions

Year: 2026 Venue: Journal of Cheminformatics Authors: Reem Alsulami, Robert Lehmann, Anuj Daga, Sumeer A. Khan, Raik Gr\"unberg, Ahmed Abogosh, David Gomez Cabrero, Stefan T. Arold, Robert Hoehndorf, Jesper Tegner, Narsis A. Kiani DOI: 10.1186/s13321-026-01167-9 Abstract INTRODUCTION: Predicting drug-target interactions remains a significant challenge in drug development and lead optimization. Recent advances have leveraged machine learning algorithms to model drug-target interactions from molecular and sequence data. MATERIALS AND METHODS: In this work, we use Evolutionary Scale Modeling (ESM

Neuro-symbolic representation learning on biological knowledge graphs

Year: 2017 Venue: Bioinformatics Authors: Mona Alshahrani, Mohammad Asif Khan, Omar Maddouri, Akira R. Kinjo, Nuria Queralt-Rosinach, Robert Hoehndorf Abstract Motivation: Biological data and knowledge bases increasingly rely on Semantic Web technologies and the use of knowledge graphs for data integration, retrieval and federated queries. In the past years, feature learning methods that are applicable to graph-structured data are becoming available, but have not yet widely been applied and evaluated on structured biological knowledge. Results: We develop a novel method for feature learning on

Combining biomedical knowledge graphs and text to improve predictions for drug-target interactions and drug-indications

Year: 2022 Venue: PeerJ Authors: Mona Alshahrani, Abdullah Almansour, Asma Alkhaldi, Maha A. Thafar, Mahmut Uludag, Magbubah Essack, Robert Hoehndorf DOI: 10.7717/peerj.13061 Abstract Biomedical knowledge is represented in structured databases and published in biomedical literature, and different computational approaches have been developed to exploit each type of information in predictive models. However, the information in structured databases and literature is often complementary. We developed a machine learning method that combines information from literature and databases to predict drug

A Machine Learning Based Approach for Similarity Search on Biodiversity Knowledge Graphs

Year: 2019 Venue: Biodiversity Information Science and Standards Authors: Claus Weiland, Maxat Kulmanov, Marco Schmidt, Robert Hoehndorf Abstract Mass biodiversity data from scientific collections will be provided by world-wide digitization efforts like iDigBio in the U.S and DiSSCo in Europe. This opens up an increasing amount of data on wild type organisms, which enables the building of large biodiversity knowledge graphs comprising, inter alia, sequence, trait and occurrence data. Knowledge graphs model information in the form of entities and their relationships expressed in good practice

DeepPheno: Predicting single gene loss-of-function phenotypes using an ontology-aware hierarchical classifier

Year: 2020 Venue: PLOS Computational Biology Authors: Maxat Kulmanov, Robert Hoehndorf DOI: 10.1371/journal.pcbi.1008453 Abstract Author summary Gene–phenotype associations can help to understand the underlying mechanisms of many genetic diseases. However, experimental identification, often involving animal models, is time consuming and expensive. Computational methods that predict gene–phenotype associations can be used instead. We developed DeepPheno, a novel approach for predicting the phenotypes resulting from a loss of function of a single gene. We use gene functions and gene expression

DeepGOPlus: improved protein function prediction from sequence

Year: 2020 Venue: Bioinformatics Authors: Maxat Kulmanov, Robert Hoehndorf Abstract Protein function prediction is one of the major tasks of bioinformatics that can help in wide range of biological problems such as understanding disease mechanisms or finding drug targets. Many methods are available for predicting protein functions from sequence based features, protein–protein interaction networks, protein structure or literature. However, other than sequence, most of the features are difficult to obtain or not available for many proteins thereby limiting their scope. Furthermore, the

Predicting protein functions using positive-unlabeled ranking with ontology-based priors

Year: 2024 Venue: Bioinformatics Authors: Fernando Zhapa-Camacho, Zhenwei Tang, Maxat Kulmanov, Robert Hoehndorf DOI: 10.1093/bioinformatics/btae237 Abstract Automated protein function prediction is a crucial and widely studied problem in bioinformatics. Computationally, protein function is a multilabel classification problem where only positive samples are defined and there is a large number of unlabeled annotations. Most existing methods rely on the assumption that the unlabeled set of protein function annotations are negatives, inducing the false negative issue, where potential positive

Semantic similarity and machine learning with ontologies

Year: 2020 Venue: Briefings in Bioinformatics Authors: Maxat Kulmanov, Fatima Zohra Smaili, Xin Gao, Robert Hoehndorf DOI: 10.1093/bib/bbaa199 Abstract Ontologies have long been employed in the life sciences to formally represent and reason over domain knowledge and they are employed in almost every major biological database. Recently, ontologies are increasingly being used to provide background knowledge in similarity-based analysis and machine learning models. The methods employed to combine ontologies and machine learning are still novel and actively being developed. We provide an overview
Schematic of machine learning for survival analysis: a neural network feeding stratified descending survival curves

Mohammad Ashhad successfully defends MSc thesis on machine learning for survival analysis

1 min read ·
Congratulations to Mohammad Ashhad, who successfully defended his Master’s thesis “ Machine Learning Methods for Survival Analysis: From Outcome-Conditioned Data Synthesis to Decoupled Ranking and Calibration” on 3 May 2026 in KAUST’s Bioengineering Program. Mohammad will now continue in the group as a PhD student in the Computer Science Program. The thesis tackles two persistent obstacles to building accurate, generalizable survival models from biomedical data: the scarcity of shareable training data, and the trade-off between expressiveness and robustness that classical and deep-learning

Neuro-symbolic AI

Neuro-symbolic methods in bioinformatics aim to combine the deductive guarantees of symbolic knowledge with the inductive power of statistical learning. Our group develops methods that map entities described in formal ontologies into vector spaces while preserving the semantic relations expressed by their axioms, so that downstream models can use background knowledge directly in similarity search, link prediction, and classification. The distinctive angle at KAUST is a focus on description logics as the source of structure: we design embedding constructions for languages such as EL++ and ALC

Genomic Context

Bacterial protein-function prediction that exploits operon and genome-neighbourhood structure in addition to sequence and homology. Get it GitHub: https://github.com/bio-ontology-research-group/Genomic_context ★ 3 Developed in projects Computational methods for functional metagenomics: from protein functions to multi-scale interactions Category: Protein Function Prediction

PhenoGoCon

Predicts gene–phenotype associations from predicted Gene Ontology functions; bridges GO function prediction and HPO/MPO phenotype prediction. Get it GitHub: https://github.com/bio-ontology-research-group/phenogocon ★ 3 Developed in projects Computational methods for functional metagenomics: from protein functions to multi-scale interactions Category: Protein Function Prediction

GO-Agent

LLM-agent framework that decomposes protein-function prediction into tool-calling sub-tasks (sequence search, structure lookup, domain reasoning) and stitches the evidence into a final GO annotation. Get it GitHub: https://github.com/bio-ontology-research-group/go-agent ★ 7 Developed in projects KAUST Center of Excellence for Generative AI (Health and Wellness, BCB theme) Category: Protein Function Prediction

DeepPheno

Predicts loss-of-function organism-level phenotypes (HPO/MPO) directly from a gene's annotated functions, using a hierarchical neural classifier over phenotype ontologies. Get it GitHub: https://github.com/bio-ontology-research-group/deeppheno ★ 6 Developed in projects CompleX: Variant Prioritization in Complex Disease Category: Protein Function Prediction

PU-GO

Positive-unlabeled ranking of protein functions with ontology-based priors; directly addresses the partial-annotation problem in CAFA benchmarks. Get it GitHub: https://github.com/bio-ontology-research-group/PU-GO ★ 4 Developed in projects Computational methods for functional metagenomics: from protein functions to multi-scale interactions Category: Protein Function Prediction

DeepGOMeta

DeepGO trained specifically for metagenomic communities; predicts functional roles of proteins recovered from environmental samples and links them to biogeochemical processes. Get it GitHub: https://github.com/bio-ontology-research-group/deepgometa ★ 8 Developed in projects Enabling desert revegetation by AI-tailored soil microbiome fortification Enabling mangrove restoration by AI-tailored microbiome fortification Category: Protein Function Prediction

DeepGOZero

Zero-shot extension of DeepGO using model-theoretic ELEmbeddings to predict GO classes that have never been observed during training. Get it GitHub: https://github.com/bio-ontology-research-group/deepgozero ★ 34 Developed in projects Computational methods for functional metagenomics: from protein functions to multi-scale interactions Category: Protein Function Prediction

DeepGO2

Next-generation DeepGO model with transformer protein embeddings and improved hierarchical multi-label prediction. Get it GitHub: https://github.com/bio-ontology-research-group/deepgo2 ★ 57 Developed in projects Computational methods for functional metagenomics: from protein functions to multi-scale interactions Category: Protein Function Prediction

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. Get it GitHub: https://github.com/bio-ontology-research-group/deepgoplus ★ 103 Homepage: https://deepgo.cbrc.kaust.edu.sa/deepgo/ Developed in projects Computational methods for functional metagenomics: from protein functions to multi-scale interactions Bio2Vec: Smart analytics infrastructure for the life sciences Category: Protein Function Prediction

Interpretable Learning

Generates interpretable symbolic rules from learned representations over biomedical knowledge bases. Get it GitHub: https://github.com/bio-ontology-research-group/interpretable-learning ★ 5 Developed in projects Towards sound, complete, and explainable machine learning with biomedical ontologies (CRG11) Category: Ontology Embedding & Machine Learning

EL2Box

Box-shaped geometric embeddings for EL++ that strengthen the topological guarantees of EL Embeddings. Get it GitHub: https://github.com/bio-ontology-research-group/EL2Box_embedding ★ 2 Developed in projects IBNSINA-QI: Integrating Biomedical Networks and Semantic Information for Neural network Analysis of Quantitative Information Category: Ontology Embedding & Machine Learning

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*. Get it GitHub: https://github.com/bio-ontology-research-group/walking-rdf-and-owl ★ 47 Developed in projects Bio2Vec: Smart analytics infrastructure for the life sciences Category: Ontology Embedding & Machine Learning

DL2Vec

Encodes description-logic axioms as a directed graph and learns embeddings via random walks; widely used for downstream gene-disease and protein-function prediction. Get it GitHub: https://github.com/bio-ontology-research-group/DL2Vec ★ 20 Developed in projects Bio2Vec: Smart analytics infrastructure for the life sciences Category: Ontology Embedding & Machine Learning

Onto2Vec

Representation learning for ontologies and their annotations by treating logical axioms as natural-language sentences; predecessor of OPA2Vec. Get it GitHub: https://github.com/bio-ontology-research-group/onto2vec ★ 21 Developed in projects Bio2Vec: Smart analytics infrastructure for the life sciences Category: Ontology Embedding & Machine Learning

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. Get it GitHub: https://github.com/bio-ontology-research-group/opa2vec ★ 37 Developed in projects Bio2Vec: Smart analytics infrastructure for the life sciences CompleX: Variant Prioritization in Complex Disease Category: Ontology Embedding & Machine Learning

DELE

Deductive EL embeddings: enrich training data with the deductive closure of an ontology before learning, so embeddings recover entailment rather than only asserted axioms. Get it GitHub: https://github.com/bio-ontology-research-group/DELE ★ 2 Developed in projects IBNSINA-QI: Integrating Biomedical Networks and Semantic Information for Neural network Analysis of Quantitative Information Category: Ontology Embedding & Machine Learning

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. Get it GitHub: https://github.com/bio-ontology-research-group/el-embeddings ★ 28 Developed in projects IBNSINA-QI: Integrating Biomedical Networks and Semantic Information for Neural network Analysis of Quantitative Information Category: Ontology Embedding & Machine 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. Get it GitHub: https://github.com/bio-ontology-research-group/mowl ★ 91 Homepage: https://mowl.readthedocs.io Developed in projects IBNSINA-QI: Integrating Biomedical Networks and Semantic Information for Neural network Analysis of Quantitative Information Towards sound, complete, and explainable machine learning with biomedical ontologies

Mahdi Bu Ali

M.S. (former), Computer Science
Mahdi Bu Ali completed his MSc in Computer Science at KAUST in 2025 under the supervision of Robert Hoehndorf. His research lay at the intersection of formal mathematics, automated theorem proving, and large language models. His thesis, Automated Theorem Proving with Large Language Models in Lean: An Exploration of Specialized In-Context Learning and General-Purpose Hierarchical Architectures, addresses the long-standing goal of automating mathematical reasoning in interactive theorem provers such as Lean. Automated theorem proving is constrained both by the cost of exploring vast proof search

Safana Bakheet

M.S. (former), Bioengineering
Safana Bakheet completed her MSc in Bioengineering at KAUST in 2025 under the supervision of Robert Hoehndorf. Her work was carried out in collaboration with Fernando Zhapa-Camacho and developed methods for prioritising disease-causing genes by combining ontology-based phenotype representations with supervised knowledge graph embeddings. Her thesis, An Inductive, Supervised Gene-Disease Associations Method, introduces a ranking model that scores candidate disease-causing genes based on the similarity of their associated phenotypes, represented in the Mammalian Phenotype Ontology (MP) and the

Application of AI in Bioinformatics (2025)

Year: 2025 Code: CS 321 Program: Computer Science Institution: KAUST Role: Instructor Topics: Biomedical informatics, Neuro-symbolic AI, Protein function

Introduction to Artificial Intelligence (2019)

Year: 2019 Code: CS 290 E Program: Computer Science Institution: KAUST Role: Instructor Topics: Neuro-symbolic AI

Knowledge Representation and Reasoning (2020)

Year: 2020 Code: CS 213 Program: Computer Science Institution: KAUST Role: Instructor Topics: Ontology engineering, Applied Ontology, Neuro-symbolic AI

Knowledge Representation and Reasoning (2021)

Year: 2021 Code: CS 213 Program: Computer Science Institution: KAUST Role: Instructor Topics: Ontology engineering, Applied Ontology, Neuro-symbolic AI

Data Analytics (2021)

Year: 2021 Code: CS 220 Program: Computer Science Institution: KAUST Role: Instructor Topics: Biomedical informatics, Neuro-symbolic AI

Knowledge Representation and Reasoning (2022)

Year: 2022 Code: CS 213 Program: Computer Science Institution: KAUST Role: Instructor Topics: Ontology engineering, Applied Ontology, Neuro-symbolic AI

Data Analytics (2023)

Year: 2023 Code: CS 220 Program: Computer Science Institution: KAUST Role: Instructor Topics: Biomedical informatics, Neuro-symbolic AI

Neurosymbolic AI (2024)

Year: 2024 Code: CS 394D Program: Computer Science Institution: KAUST Role: Instructor Topics: Neuro-symbolic AI, Applied Ontology

Knowledge Representation and Reasoning (2024)

Year: 2024 Code: CS 213 Program: Computer Science Institution: KAUST Role: Instructor Topics: Ontology engineering, Applied Ontology, Neuro-symbolic AI

Neurosymbolic AI (2026)

Year: 2026 Code: CS 394D Program: Computer Science Institution: KAUST Role: Instructor Topics: Neuro-symbolic AI, Applied Ontology

Knowledge Representation and Reasoning (2026)

Year: 2026 Code: CS 213 Program: Computer Science Institution: KAUST Role: Instructor Topics: Ontology engineering, Applied Ontology, Neuro-symbolic AI
Data integration and ontologies for microbial cell factories — research illustration

Data integration and ontologies for microbial cell factories

-
Industrial biotechnology depends on integrating data from sources that were never designed to be combined: enzyme and reaction databases, protein-family resources, pathway maps, strain collections, phenotypic screens, and the experimental literature itself. Each uses different identifiers, different vocabularies, and different levels of granularity, and naive integration produces brittle pipelines that break whenever a source is updated. The Microbial Cell Factories initiative at KAUST's Computational Bioscience Research Center (CBRC, now dissolved), led by Vladimir Bajic with Hoehndorf as CoI
Bio2Vec: Smart analytics infrastructure for the life sciences — research illustration

Bio2Vec: Smart analytics infrastructure for the life sciences

-
By the mid-2010s the life sciences had produced an extraordinary investment in machine-readable knowledge: biomedical ontologies were used throughout biology to annotate data, and large RDF knowledge graphs such as Bio2RDF aggregated billions of statements from dozens of major databases. At the same time, large personal genomic datasets, the UK 100,000 Genomes project, UK Biobank, and the Saudi Human Genome Program, were coming online, and translating these into clinical insight depended on integrating them with that existing background knowledge. Generic knowledge-graph machine learning
Improvement of genetic variant prioritization technology — research illustration

Improvement of genetic variant prioritization technology

-
By 2018 the PhenomeNET Variant Predictor (PVP) had been validated as a research tool for prioritizing causative variants in Mendelian disease, and a clear question had emerged: could the same phenotype-driven, knowledge-graph-based machinery be retargeted to cancer, where the variant landscape is somatic, heterogeneous, and dominated by combinations of driver and passenger mutations rather than single pathogenic alleles? This one-year KAUST Center Partnership Fund project (2018–2019, OSR-2018-CPF-3657-0; with Schofield in Cambridge, Gkoutos in Birmingham, and Bajic at KAUST) was the
Improving health of Saudi population — research illustration

Improving health of Saudi population

-
Translating modern biomedical knowledge into healthcare for Saudi Arabia requires methods that work on local data: a population with high consanguinity, a distinctive spectrum of inborn errors of metabolism, and clinical phenotypes that are not always well represented in international reference resources. The project, led by Hoehndorf at KAUST's Computational Bioscience Research Center (CBRC, now dissolved), developed health-informatics methods and resources oriented towards Saudi-population data, with a focus on rare disease, drug treatment, and the formal representation of phenotype
CompleX: Variant Prioritization in Complex Disease — research illustration

CompleX: Variant Prioritization in Complex Disease

-
The hardest cases in clinical genome sequencing are the ones where no single variant explains the disease. As Mendelian gene discovery slows and the diagnostic rate for whole-exome sequencing stalls below 50%, growing evidence points to oligogenic and polygenic origins: combinations of medium-rare or common alleles that, individually, look unremarkable. Population-level approaches lack the power to find them, and traditional single-gene Mendelian reasoning ignores them. The CompleX project (2019–2021, with the Universities of Cambridge and Birmingham) set out to break this impasse by extending
Development of Algorithms for Biotechnology and Biomedical Applications — research illustration

Development of Algorithms for Biotechnology and Biomedical Applications

-
This CBRC Competitive Funding project (2021–2023, run under the now-dissolved Computational Bioscience Research Center) functioned as an umbrella program for algorithm development across the group's biotechnology and biomedical work, with a deliberate focus on metabolic modeling — predicting metabolic function from genome, predicting interactions from structure, and learning from interaction networks to identify disease-relevant biology. The unifying scientific bet was that systems biology requires algorithms that operate not on isolated molecules but on the networks of interactions within and
IBNSINA-QI: Integrating Biomedical Networks and Semantic Information for Neural network Analysis of Quantitative Information — research illustration

IBNSINA-QI: Integrating Biomedical Networks and Semantic Information for Neural network Analysis of Quantitative Information

-
Biological measurements are inherently high-dimensional and heterogeneous: omics platforms produce thousands to millions of features per individual, and they coexist with qualitative information such as diagnoses, phenotype calls, and prescriptions. Biomedical ontologies and knowledge graphs encode rich qualitative background knowledge, but they are largely disconnected from the quantitative measurements that gave rise to the categorical phenotypes in the first place. Conversely, graph neural networks and other methods that handle quantitative data on graphs do not yet exploit the formal
Computational methods for functional metagenomics: from protein functions to multi-scale interactions — research illustration

Computational methods for functional metagenomics: from protein functions to multi-scale interactions

-
Metagenomic sequencing has made it routine to read the DNA of an entire microbial community, but most analysis pipelines stop at taxonomic composition or at the level of individual protein families. The really biologically informative questions, which proteins do what, which proteins interact, which metabolic pathways are reconstructible, and how the community as a whole interacts with its environment or host, remain largely out of reach computationally. Even associations that are very robust empirically, for example between gut microbiome composition and colorectal cancer or inflammatory
Enabling mangrove restoration by AI-tailored microbiome fortification — research illustration

Enabling mangrove restoration by AI-tailored microbiome fortification

-
This short collaboration (2023, funded under the KAUST Center for Living Innovation program) developed AI-tailored microbiome work in support of coastal mangrove restoration in Saudi Arabia. The project ran as a sister activity to the BORG group's larger desert-microbiome program with the NTGC, applying analogous metagenomic and functional-annotation pipelines to mangrove sediment and root-associated communities rather than to arid-soil systems. Heribert Hirt (DARWIN21, KAUST) was the Principal Investigator, with Hoehndorf as Co-Investigator providing the computational microbiome analysis
Enabling desert revegetation by AI-tailored soil microbiome fortification — research illustration

Enabling desert revegetation by AI-tailored soil microbiome fortification

-
Roughly twelve million hectares of dryland are lost every year worldwide, and the megaprojects launched to reverse desertification — the Sahel's Great Green Wall is the canonical example — typically fail because most planted trees die when irrigation stops. The biological reason is that desert soils are low in nutrients and high in salinity, and the microbial communities that mediate nitrogen, phosphate, and mineral acquisition for plants are missing or unbalanced. This project, led by Heribert Hirt's plant-microbiome group with our group and the Modeling and Simulation group of Gabriel Wittum
Disease Models from Patient-derived Leukemic Cells in Biomimetic Peptide Scaffolds for Precision Medicine Applications — research illustration

Disease Models from Patient-derived Leukemic Cells in Biomimetic Peptide Scaffolds for Precision Medicine Applications

-
Acute myeloid leukemia (AML) has a five-year survival rate of roughly 24% in Saudi Arabia and is the second most common adult leukemia subtype in the Kingdom. Treatment selection still rests on cytogenetic risk stratification — a coarse instrument that misses much of the molecular heterogeneity now known to drive drug response. Compounding the problem, almost every published AML drug-response prediction model has been trained on Western cohorts, and existing 3D culture systems for leukemia rely on animal-derived matrices that resist standardization. This project (2023–2026, KAUST Smart-Health
Towards sound, complete, and explainable machine learning with biomedical ontologies (CRG11) — research illustration

Towards sound, complete, and explainable machine learning with biomedical ontologies (CRG11)

-
Deep learning has driven much of the recent progress in bioinformatics, but the resulting models are essentially black boxes: it is hard or impossible to ask why a prediction was made, to verify it against existing knowledge, or to detect biases that emerge from the interaction of dataset, architecture, and training objective. In clinical and biomedical settings these limitations matter, both because clinicians need to trust the systems they use and because formal guarantees of soundness and fairness can only be obtained when the inferential machinery can be inspected. The grand challenge
KAUST Center of Excellence for Generative AI (Health and Wellness, BCB theme) — research illustration

KAUST Center of Excellence for Generative AI (Health and Wellness, BCB theme)

The KAUST Center of Excellence for Generative AI has a Health and Wellness theme, and within that theme the Bioinformatics and Computational Biology (BCB) workstream is responsible for the bio-side problems that generic generative models cannot solve on their own: protein and antibody design, population-specific drug development, and clinical foundation models that can actually reason with structured biomedical knowledge. The BORG group leads the neuro-symbolic and ontology-based components of this work (Sub-WBS FCC/1/5940-07-02), in partnership with Xin Gao's group on AI/drug design and the
Personalized cancer treatment prediction (KCSH Pathway to Impact 2025) — research illustration

Personalized cancer treatment prediction (KCSH Pathway to Impact 2025)

-
Predicting which drug will work for which patient is one of the oldest unsolved problems in oncology, and it is especially acute in Saudi Arabia, where colorectal cancer is the most common cancer in men and is shifting toward younger ages, and hepatocellular carcinoma incidence has roughly doubled in two decades. Contemporary machine learning models for drug response largely ignore the decades of curated biological knowledge that surrounds them, and they are almost never trained on Saudi populations. This project, funded under the KCSH Pathway to Impact 2025 call and running 2025-2026, aims to
A public Saudi pangenome as reference for genomics in the Middle East — research illustration

A public Saudi pangenome as reference for genomics in the Middle East

-
Human reference genomes such as GRCh38 are built largely from a single individual of mixed African-European ancestry, and they systematically misrepresent the genetic structure of populations that diverged from that backbone, especially populations like those of the Arabian Peninsula that were historically organized in tribal structures with high in-tribe marriage. The consequence in clinical practice is a reference bias that inflates the apparent variant count by millions, complicates interpretation, and lengthens diagnostic turnaround. With more than half of marriages in Saudi Arabia

Robert Hoehndorf

Associate Professor, Computer Science
Robert Hoehndorf is an Associate Professor of Computer Science at KAUST. His research focuses on bioinformatics, knowledge representation and reasoning, Semantic Web technologies, and neuro-symbolic methods.

Olga Mashkova

Ph.D. Student, Computer Science
PhD student in Computer Science. Research interests: machine learning, bioinformatics, mathematics, knowledge representation and reasoning, deep learning, algorithms.

Fernando Patricio Zhapa-Camacho

Postdoctoral Research Fellow, Computer Science
Fernando Patricio Zhapa Camacho is an M.S./Ph.D. candidate in the KAUST Bio-Ontology research group under the supervision of Professor Robert Hoehndorf. Before joining KAUST, Fernando obtained a bachelor's degree in Information Technology from Yachay Tech University, Ecuador. Research Interests Fernando’s research interests include knowledge representation and reasoning, machine learning and bioinformatics. He is interested in artificial intelligence and logic integration with applications in biology. Professional Profile: • 2020 to present: Graduate Student, KAUST, Thuwal, Saudi Arabia. •