Setting up cloudgene instance for polygenic scoring

[fronay]

Thanks for the brand new version of imputationserver!
How does one set this up locally to execute / enable the polygenic scoring pipeline on top of imputation, if using the ./cloudgene server command?

(I realise one can enable pgs in the nextflow.config file, but presumably there are other settings to tick + one would also have to download the catalogue to the correct reference path, not clear to me from the documentation what best practices are intended.)

[seppinho]

Hi. Setting up a local PGS server is not well documented yet. Will start working on that. Which reference panel are you planning to use?
Best.
Sebastian

[fronay]

Dear Sebastian, apologies for the late reply. I was planning on using the 1000Genomes Phase 3 v5 reference panel as an established standard - but would be open to switching to different reference panels (say, HRC for Europeans) if more widely recommended as best practices in PGS literature. (I noted for instance that Privé et al, 2022, who seem to be a source for many useful scores in the PGS catalog repository, mentioned use of HapMap 3 for this purpose, which I believe isn't available as a standard choice on MIS).

Thanks for considering adding some documentation for the PGS server option!

[seppinho]

For genotype imputation, 1000G setup is already available: https://genepi.github.io/michigan-imputationserver/tutorials/local-server/.

[fronay]

Thanks, yes, local imputation with 1000G works! Just regarding running the local server PGS scoring option at all - independent of reference panel - I look forward to any extra pointers in documentation (if it's something you were considering working on anyway.)

[seppinho]

Great to hear that works. For PGS, we just need to prepare the PGS data for download.