The finemap app performs statistical fine-mapping of a genomic locus using SuSiE-RSS. It uses a region-subset, harmonised GWAS summary-statistics file and the corresponding study ID to estimate which variants are most likely to explain the association signal.
The app returns a downloadable archive containing:
- Posterior inclusion probabilities (PIPs) for individual variants
- Credible sets of variants that are likely to contain the causal signal
Use finemap after identifying a GWAS locus of interest, when you want to prioritise candidate causal variants for follow-up.
Before you begin
You will need:
- A harmonised GWAS summary-statistics file restricted to the locus of interest
- The corresponding Sequoia study ID
- A project in which to save the job and output files
Name your job
Enter a descriptive name in the Prefix field. This is required.
[Screenshot 1: Red box around the Prefix field.]
For example:
CAD_9p21_finemapLDL_APOE_SuSiEBMI_FTO_locus
Provide regional summary statistics
Upload or select the regional harmonised GWAS summary-statistics file in the sumstat section.
[Screenshot 2: Highlight the sumstat section, including Upload New, Select Existing, and From Jobs Output.]
You can provide the file in three ways:
- Upload New — Upload a file from your computer.
- Select Existing — Choose a file already stored in the portal.
- From Jobs Output — Use a file generated by a previous Sequoia job.
Supported file formats are .tsv and .gz.
[Screenshot 3: Red box around the upload area and the supported file formats.]
The file should contain summary statistics for the locus being fine-mapped, rather than an unrestricted genome-wide dataset.
Enter the study ID
Enter the Sequoia identifier for the study that produced the uploaded summary-statistics file in studyid.
[Screenshot 4: Red box around the studyid field.]
The study ID allows the app to retrieve the relevant study information and apply the fine-mapping workflow correctly.
Set the maximum number of iterations
Use niter to set the maximum number of iterations for the SuSiE-RSS algorithm. The default value is 100.
[Screenshot 5: Red box around the niter field.]
In most cases, the default should be retained. Increase this value only if advised by your analysis plan or if the model does not converge with the default setting.
Review the quality-control settings
The geno, mind, and mac fields define quality-control thresholds used in the fine-mapping workflow.
[Screenshot 6: Highlight the geno, mind, and mac fields.]
- geno — Variant-level missingness threshold. Default:
0.01 - mind — Sample-level missingness threshold. Default:
0.01 - mac — Minimum minor allele count threshold. Default:
6
These defaults are appropriate for routine use. Change them only when you have a predefined quality-control protocol.
Select a project
Choose the destination project in the project_id dropdown.
[Screenshot 7: Red box around the project_id dropdown and the Research Hub → Projects note.]
If necessary, create a project in Research Hub → Projects before launching the app.
Launch the fine-mapping analysis
Review the summary-statistics file, study ID, and settings, then select Launch App.
[Screenshot 8: Red box and arrow pointing to “Launch App”.]
When the analysis is complete, download the output archive from the selected project or the job results area.
Interpreting the output
- Posterior inclusion probability (PIP) indicates how strongly the data support a variant being included in the model for the association signal. Higher PIP values indicate stronger support.
- A credible set is a group of variants that jointly contains the likely causal variant with a defined level of confidence.
Fine-mapping prioritises variants statistically; it does not by itself establish biological function or causality. Candidate variants should be evaluated alongside functional annotation, eQTL evidence, colocalisation results, and experimental data.
Tips
- Use a regional, harmonised summary-statistics file that captures the full association signal and nearby correlated variants.
- Confirm that the entered study ID corresponds exactly to the supplied summary-statistics file.
- Retain the default QC and iteration settings unless you have an analysis-specific reason to change them.
- Use the PIP and credible-set outputs to prioritise variants for downstream functional or target-discovery work.