Until recently, we performed all fine-mapping, but only for studies with summary statistics and with primarily European ancestry, using an LD reference panel from 10,000 randomly sampled UK biobank individuals.
We use GCTA-cojo to detect independent genome-wide significant signals, conditioning on all other genome-wide significant signals within 2 Mb. We use the standard coloc software from Chris Wallace’s lab on the conditionally independent summary statistics from GCTA–cojo-cond.
In version 5 of the Genetics Portal (June 2021 release), we also integrated fine-mapping performed by FinnGen, which used an ancestry-matched whole genome sequencing-based reference panel and modern fine-mapping methods (SuSiE).
Currently integrating fine-mapping in this way precludes us from doing colocalisation tests for secondary signals in these other datasets (FinnGen). However, we have been engaging with FinnGen and the eQTL Catalogue to make available their raw fine-mapping output, which would enable us to both use their high-quality fine-mapping while also doing colocalisation with secondary signals.
For more information on our fine mapping and colocalisation pipelines, check out the Genetics Portal documentation:
- Fine mapping: Assigning Variants to Disease (V2D) - Open Targets Genetics Documentation
- Colocalisation: Colocalisation analysis - Open Targets Genetics Documentation
And finally, for more information on the technology underpinning the Portal, see: What Technologies Do We Use? - Open Targets Genetics Documentation