Run an Existing Loci List¶
Use this tutorial when you already have locus-level files. This is the shortest path into CREDTOOLS.
The Required Files¶
For each row in your loci list, CREDTOOLS reads files by prefix:
{prefix}.sumstats.gz # or {prefix}.sumstat
{prefix}.ld.npz # or {prefix}.ld
{prefix}.ldmap # or {prefix}.ldmap.gz
Your loci_list.txt should be tab-separated:
locus_id chr start end popu cohort sample_size prefix
locus_9p21 9 21900000 22100000 EUR UKBB 400000 data/EUR_UKBB_9p21
locus_9p21 9 21900000 22100000 AFR MVP 90000 data/AFR_MVP_9p21
All rows with the same locus_id must use the same chr, start, and end.
Do not include file extensions in prefix
Use data/EUR_UKBB_9p21, not data/EUR_UKBB_9p21.sumstats.gz.
Validate the Shape¶
Before a large run, inspect a few rows:
Then ask CREDTOOLS for command help:
Run the Pipeline¶
This creates one output directory per locus:
results/
- locus_9p21/
- pips.txt.gz
- credible_sets_summary.txt.gz
- causal_variants.txt.gz
- run_summary.log
- overall_run_summary.log
Choose the Meta Method¶
Use --meta-method to control how rows are combined before fine-mapping:
Check the Main Result¶
Read the PIP table:
Read the credible set summary:
If no credible set is reported, check QC and the smallest p-values in the locus. No credible set can be a valid result when the evidence is weak.