Why Multi-Omics for Immune-Mediated Disorders

Immune-mediated diseases look like single diagnoses but behave like populations of distinct molecular diseases. Two patients who both carry a diagnosis of rheumatoid arthritis, lupus, or inflammatory bowel disease can have inflammation driven by entirely different cell types, cytokine circuits, and genetic risk — which is why a drug that transforms one patient does little for the next.

The central problem of the field is not describing that these diseases are heterogeneous; it is resolving which axis of heterogeneity matters for a given patient and a given therapy. No single measurement captures that. Genetics tells you inherited risk but not the current state of the tissue. Bulk transcriptomics averages over the very cell types whose balance defines the disease. Protein abundance — the layer closest to drug action — correlates only modestly with mRNA (on the order of r ≈ 0.4 in many tissues), so a transcript-only readout misses most of the biology a therapeutic will actually engage. Each layer is a projection; the disease lives in their intersection.

The modern multi-omics playbook

Multi-omics integration is the discipline of reconstructing that intersection, and the approach is now consistent across disorders: genome-wide association studies map inherited risk; functional genomics (expression QTLs, single-cell chromatin accessibility, colocalization) converts non-coding risk loci into the specific genes and cell types that carry them; single-cell and spatial transcriptomics identify the pathogenic cell states in the affected tissue; and bulk molecular profiling stratifies patients for treatment-response prediction. Together these turn a clinical label into a mechanistic, cell-resolved, patient-specific picture.

Where it already pays off

The payoff is tangible where the data are rich. In inflammatory bowel disease, cellular modules identified by single-cell analysis of mucosa are associated with resistance to anti-TNF therapy (Martin et al., Cell 2019; Smillie et al., Cell 2019). In psoriasis, convergent genetic, transcriptomic, and functional evidence pinned the IL-23/IL-17 axis — and that convergence translated directly into a generation of successful biologics (Tsoi et al., Nat Genet 2012; Reynolds et al., Science 2021). These are proofs that integrated multi-omics is not description for its own sake; done well, it nominates targets and stratifies patients.

"The disease is not in any one layer. It is in the agreement — and the disagreement — between them."

A platform that carries a scientist from raw multi-omics data to an integrated, quality-checked, interpretable result — without a bioinformatics team per project — is what makes this routine rather than heroic. The rest of this series walks through the public data that makes it possible, the methods that integrate it, and where the field can and cannot yet deliver, disorder by disorder.