A crucial component of the treatment of genetic disorders is identifying and characterising the genes and gene modules that drive disease processes. Recent advances in Next-Generation Sequencing improve the prospects for achieving this goal. However, many machine learning techniques are not explainable and fail to account for gene correlations. In this work, we develop a comprehensive set of explainable machine learning techniques to perform patient stratification for inflammatory bowel disease. We focus on Crohn's disease (CD) and its subtypes: CD with deep ulcer, CD without deep ulcer and IBD-controls. We produce an interpretable probabilistic model over disease subtypes using Gaussian Mixture Modelling. We then apply class-contrastive and feature-attribution techniques to identify potential target genes and modules. We modify the widely used kernelSHAP (Shapley Additive Explanations) algorithm to account for gene correlations. We obtain relevant gene modules for each disease subtype. We develop a class-contrastive technique to visually explain why a particular patient is predicted to have a particular subtype of the disease. We show that our results are relevant to the disease through Gene Ontology enrichment analysis and a review of the literature. We also uncover some novel findings, including currently uncharacterised genes. These approaches maybe beneficial, in personalised medicine, to inform decision-making regarding the diagnosis and treatment of genetic disorders. Our approach is model-agnostic and can potentially be applied to other diseases and domains where explainability and feature correlations are important.

The authors have declared no competing interest.

We would like to express our gratitude to the Cambridge Trust and Google DeepMind for their generous support through the DeepMind Cambridge Scholarship, awarded to SO for her postgraduate studies at the University of Cambridge. SB acknowledges funding from the Accelerate Programme for Scientific Discovery Research Fellowship. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The views expressed are those of the authors and not necessarily those of the funders.

I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained.

Yes

The details of the IRB/oversight body that provided approval or exemption for the research described are given below:

The study used only openly available human data that were originally located at: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE57945

I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals.

Yes

I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance).

Yes

I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable.

Yes