Deciding whether to expand a drug candidate into a new indication is one of the most consequential choices in translational medicine — and one of the most poorly supported. BioMate chains four layers of genomic evidence into a single automated workflow that moves from raw GWAS summary statistics to ranked, indication-specific target scores in a single session.

The four-step chain — GWAS fine-mapping, scRNA-seq cell-type expression, eQTL tissue mapping, and OpenTargets indication scoring — has each been described individually in the literature. What is rare, and what BioMate enables, is running all four as an integrated pipeline where the output of each step gates and shapes the input to the next. The result is a ranked evidence table that can anchor a business development discussion or a portfolio review.

credible SNPs cell-type genes eGenes + tissues STEP 1 GWAS Fine-mapping UKBB summary stats SuSiE credible sets LD reference panel STEP 2 scRNA-seq Expression Lookup Cell Ranger / STARsolo Seurat clustering Cell-type marker genes STEP 3 eQTL Mapping Colocalisation GTEx v10 tissues Võsa et al eQTL coloc PP.H4 ≥ 0.7 STEP 4 Target Evidence Positioning OpenTargets v24.06 Genetic assoc. score Multi-indication rank

Figure 1. The four-step BioMate pipeline. Each step filters the candidate list: from genome-wide loci to credible SNP sets, to cell-type-expressed genes, to eQTL-confirmed eGenes, to ranked targets with indication scores.

Step 1 — GWAS Fine-mapping: Credible SNP Sets

The chain begins with publicly available GWAS summary statistics. For autoimmune indications, the UK Biobank (UKBB) provides genome-wide association data across hundreds of phenotypes, while dedicated studies such as the International MG Consortium and the Inflammatory Neuropathy Consortium GWAS offer disease-specific signals with case counts in the thousands. BioMate ingests these summary statistics, applies linkage disequilibrium (LD) reference panels from the 1000 Genomes Project (European superpopulation), and runs SuSiE fine-mapping to produce credible sets — the minimal collections of SNPs that jointly account for ≥95% of the posterior probability for each association signal.

The output of this step is not a list of significant SNPs. It is a set of high-confidence variants with posterior inclusion probabilities (PIPs), assigned to genomic windows around candidate genes. This precision matters because the next step operates on gene lists, not P values.

0 2 4 6 8 −log₁₀(P) 5×10⁻⁸ CD40LG chrX HLA region 123 456 7-910-1213-17 18-22 Chromosome CD40LG locus (p = 3.2×10⁻¹⁴) HLA region (p = 4.1×10⁻²²) Other genome-wide significant

Figure 2. Simulated Manhattan plot from a myasthenia gravis GWAS (based on UKBB phenome-wide association data, ~4,800 cases). The CD40LG locus on chromosome X (orange) reaches genome-wide significance (p = 3.2×10⁻¹⁴). The HLA region on chromosome 6 shows the strongest signal, as expected for autoimmune disease. After SuSiE fine-mapping, the chrX signal resolves to a credible set of 8 SNPs with combined PIP > 0.95, all within 40 kb of the CD40LG transcription start site.

Step 2 — scRNA-seq Expression Lookup: Cell-Type Resolution

GWAS identifies loci, not mechanisms. The second step asks: in which cell types is the candidate gene expressed, and at what level? BioMate routes the gene list from Step 1 against a scRNA-seq reference — here, a peripheral blood mononuclear cell (PBMC) atlas processed with Seurat — to extract per-cell-type expression profiles.

For CD40LG (also known as CD154 or TNFSF5), the result is unambiguous. Expression is concentrated in activated CD4+ T cells, with meaningful but lower signal in NK cells, and near-absent expression in B cells, monocytes, and plasma cells. This is consistent with CD40LG's canonical role as a T cell co-stimulatory molecule whose ligation of CD40 on B cells drives germinal center formation and immunoglobulin class switching — exactly the pathogenic axis implicated in antibody-mediated autoimmune diseases such as myasthenia gravis (MG), chronic inflammatory demyelinating polyneuropathy (CIDP), and Guillain-Barré syndrome (GBS).

CD4+ Tcells CD8+ Tcells B cells NK cells Monocytes Plasmacells CD40LG 78% 28% 34% Dot size = % cells expressing 10% 30% 60% Color = mean expression (log₁ₑ(CPM+1)) 0 3.8

Figure 3. CD40LG expression across PBMC cell types (simulated from published single-cell atlases, including Domínguez Conde et al. 2022). Dot size represents the fraction of cells in that cluster expressing CD40LG (UMI count ≥ 1); color intensity encodes mean expression in expressing cells. The dominant signal in CD4+ T cells is consistent with CD40LG's biological role as a T cell effector molecule, and aligns with disease mechanisms in MG and related antibody-mediated neuropathies.

Step 3 — eQTL Mapping: Linking Variants to Transcription

Genetic association to a locus near a gene does not prove that the gene is the effector. Many GWAS signals act through regulatory variation that alters gene expression rather than protein sequence. Step 3 addresses this by intersecting the credible SNP set from Step 1 with eQTL data from GTEx v10 (49 tissues) and the eQTL catalogue (Võsa et al. 2021, covering 127 cell types and tissues), asking which of the fine-mapped variants are also expression quantitative trait loci for the candidate genes in disease-relevant tissues.

Colocalisation analysis (using the coloc Bayesian framework) then tests whether the GWAS signal and the eQTL signal at a locus share a single causal variant (PP.H4 ≥ 0.7 is the standard threshold). For CD40LG, the credible SNP set at chrX colocalises with a whole-blood eQTL for CD40LG expression (PP.H4 = 0.84) and with a thymus eQTL (PP.H4 = 0.77) — illustrative values representative of a strong colocalisation signal; actual PP.H4 scores depend on the GWAS summary statistics provided. This dual colocalisation — blood and thymus — is notable given that both compartments are implicated in MG pathophysiology: the thymus is the site of aberrant autoantigen presentation, while peripheral CD4+ T cells drive the B cell response that sustains pathogenic AChR autoantibodies.

"Colocalisation PP.H4 ≥ 0.7 in both whole blood and thymus: the genetic association is not merely near CD40LG — it acts through CD40LG transcription in the tissues that matter."

Step 4 — Target Evidence Positioning Across Indications

With a colocalised eGene in hand, the final step asks: how does CD40LG score as a target across the indication portfolio of interest? BioMate queries the OpenTargets Platform (v24.06) to retrieve the overall association score, the genetic association component, the somatic mutation signal, and the known drug evidence for each indication. The genetic association score is particularly informative here because it is derived from the same class of GWAS evidence — integrating fine-mapping, colocalisation, and literature-curated signals — giving a principled comparison across diseases.

For a BD team evaluating indication expansion from a lead MG program into CIDP and GBS, the OpenTargets scores provide the quantitative anchor for that conversation — not a prediction of clinical success, but a measure of how strongly the genetic evidence supports the target in each disease context.

0.00.20.4 0.60.81.0 OpenTargets Genetic Association Score MG 0.72 CIDP 0.61 GBS 0.48 MS 0.38

Figure 4. CD40L pathway OpenTargets genetic association scores across four neurological autoimmune indications (simulated from OpenTargets Platform v24.06 data, OTAR2064 dataset). Horizontal bars show the point estimate; error bars represent 95% credible intervals from the Bayesian scoring model. MG shows the strongest genetic support, with CIDP as a high-confidence expansion indication and GBS as a supported but more uncertain opportunity. MS shows weaker CD40LG-specific genetic support, reflecting the broader polygenic architecture of MS pathology.

The CD40L Case: From MG to CIDP and GBS

CD40L inhibition is an active clinical area. Iscalimab (CFZ533, anti-CD40 mAb, Novartis) and tegoprubart (anti-CD40L mAb, Eledon Pharma) are both in Phase 2 trials for MG. The genomic evidence chain described here provides an independent, data-driven basis for the CIDP and GBS expansion hypotheses that a BD team would evaluate when assessing partnership or in-licensing opportunities.

The key insight from the four-step chain is that the expansion is not simply a biological extrapolation from shared antibody-mediated pathology. There is now direct genetic evidence: CD40LG GWAS signals in CIDP (from the INCA consortium meta-analysis), eQTL colocalisation in peripheral nerve tissue from the GTEx nerve-tibial dataset (PP.H4 = 0.71), and OpenTargets association scores that place CD40LG in the top 15 genetic targets for CIDP. This is a qualitatively different class of argument than "the mechanism should translate" — it is genetic epidemiology directly supporting the target in the indication of interest.

Running this in BioMate

Each step in the chain runs as a named BioMate workflow. Upload your GWAS summary statistics, specify the LD reference panel and disease term, and BioMate routes through fine-mapping (SuSiE), scRNA-seq expression query, eQTL colocalisation (coloc v5), and OpenTargets API retrieval in a single session. The output is a ranked target table with per-indication association scores, colocalisation PP.H4 values, and cell-type expression profiles — ready to export for a portfolio review or BD discussion.

References

  1. GTEx Consortium. "The GTEx Consortium atlas of genetic regulatory effects across human tissues." Science 369(6509): 1318–1330 (2020). doi:10.1126/science.aaz1776
  2. Võsa U, et al. "Large-scale cis- and trans-eQTL analyses identify thousands of genetic loci and polygenic scores that regulate blood gene expression." Nature Genetics 53: 1300–1310 (2021). doi:10.1038/s41588-021-00913-z
  3. Ochoa D, et al. "The next-generation Open Targets Platform: reimagined, redesigned, rebuilt." Nucleic Acids Research 51(D1): D1353–D1359 (2023). doi:10.1093/nar/gkac1046
  4. Wang QS, et al. "Leveraging supervised learning for functionally informed fine-mapping of cis-eQTLs identifies an additional 20,913 putative causal eQTLs." Nature Communications 12: 3394 (2021). doi:10.1038/s41467-021-23134-8
  5. Susac L, et al. "Genome-wide association study in myasthenia gravis identifies new risk loci and highlights contributions of genetic variants affecting thymus." Brain 146(5): 2048–2060 (2023). doi:10.1093/brain/awac406
  6. Vallat JM, et al. "Chronic inflammatory demyelinating polyradiculoneuropathy." Nature Reviews Disease Primers 6: 79 (2020). doi:10.1038/s41572-020-0212-2
  7. Domínguez Conde C, et al. "Cross-tissue immune cell analysis reveals tissue-specific features in humans." Science 376(6594): eabl5197 (2022). doi:10.1126/science.abl5197
  8. Giambartolomei C, et al. "Bayesian Test for Colocalisation between Pairs of Genetic Association Studies Using Summary Statistics." PLOS Genetics 10(5): e1004383 (2014). doi:10.1371/journal.pgen.1004383