Integrating GWAS with single-cell data to identify pathogenic cell types and prioritise targets in inflammatory skin diseases
🧠 *Part of the Open Targets Hackathon *
Genome-wide association studies (GWAS) have identified hundreds of genetic loci associated with inflammatory skin diseases such as psoriasis and atopic dermatitis, but the pathogenic cell types underlying these signals remain largely unknown.
This project is meant to bridge this gap by integrating GWAS summary statistics with single-cell RNA-seq data from healthy and diseased skin to:
- Identify cell types enriched for genetic risk.
- Compute gene-level cell-type specificity scores.
- Prioritise therapeutic targets in their cellular context.
| Goal | Description |
|---|---|
| 🧩 Integrate GWAS + scRNA-seq | Combine genetic association data with cell-type resolved transcriptomics. |
| 🔬 Identify pathogenic cell types | Determine which cell populations drive disease risk in psoriasis and eczema. |
| 🧬 Prioritise target genes | Compute gene-level cell-type specificity scores to rank causal/effector genes. |
| 💡 Contribute to Open Targets Platform | Prototype a new “cell-type specificity” evidence layer. |
- Retrieve GWAS summary statistics for psoriasis and atopic dermatitis from the GWAS Catalog.
- Obtain credible sets from the Open Targets Platform.
- Gather public scRNA-seq datasets from healthy and diseased skin.
- Calculate Beta values using mtag_munge.py in CELLECT-LDSC. For ease, have an input file that looks like GWAS input
- Estimate cell-type expression specificity score using CELLEX.
- Perform genetic prioritisation using CELLECT-MAGMA or CELLECT-LDSC.
- List disease-relevant/etiologic cell types per condition.
- Cell-type specificity scores for prioritised targets
- Visualisation and summary reports
| Category | Tools / Libraries |
|---|---|
| Single-cell analysis | 🧫 Scanpy |
| GWAS–single-cell integration | CELLECT, CELLEX |
| Data sources | 🌐 GWAS Catalog, Open Targets Platform, Public scRNA-seq data from inflammatory skin diseases |
| Language | 🐍 Python (scanpy, Jupyter, CELLEX, CELLECT), snakemake |
| Google Cloud | Vertex AI |
- Ranked list of pathogenic cell types for psoriasis and eczema.
- Cell-type specificity scores for target prioritisation.
- A prototype Open Targets cell type specificity evidence feature integrating these results.
