Kevin O'Sullivan

University College Dublin / Microbiome Analytics

Advisor: Prof. Fergus Shanahan

Kevin O'Sullivan applies shotgun metagenomics and machine learning to stratify irritable bowel syndrome patients into clinically meaningful subtypes at the University College Dublin's APC Microbiome Ireland research centre, supervised by Prof. Fergus Shanahan. IBS diagnosis currently relies entirely on symptom-based criteria (Rome IV) that group together patients with fundamentally different gut microbiome architectures and likely different treatment requirements. Kevin has profiled the faecal metagenomes of 340 IBS patients and 120 healthy controls with Illumina HiSeq shotgun sequencing, generating species-level taxonomic and functional pathway abundance matrices from Kraken2 and HUMAnN3. An XGBoost classifier trained on a 70% training split correctly assigns patients to one of four microbiome-defined IBS subtypes with 83% held-out accuracy — substantially outperforming symptom-based classification for predicting low-FODMAP diet response at six months. Kevin has validated the four subtypes against plasma serotonin and tryptophan catabolite metabolomics and shown that subtype 2 (characterised by Akkermansia and Faecalibacterium enrichment) has a 3.4-fold higher dietary intervention response rate. His industry collaboration with Alimentiv Inc. — a CRO specialising in gut disease clinical trials — provides access to biobanked samples from four multicentre IBS treatment trials for independent validation.