SurPRSe!

• Supercomputing with Polygenic Risk Score Evaluation

• A workflow in Bash, R and python designed to allow easy processing and production of polygenic risk scores (both whole genome and set scores) on ARCCA's Raven supercomputer
• For a tutorial, debugging information and other documentation, please refer to the wiki
• If used for the creation of results to be published (one can dream...), please acknowledge John Hubert [email protected] and Valentina Escott-Price [email protected]
• I'm always looking for people to help in making this better, If you are familiar to git please fork this repository.
  • The code has a General Public license so you are free to make changes of your own just as long as you document them!
  • If you are unfamiliar with the 'git' structure but still want to help out, please email John Hubert @ [email protected] and we can sort something out!
• If you notice a bug or a fix that is required please add to the issues page or email [email protected]

SAVE-PRS

** COMING SOON**

• SurPRSe! Analysis Viewing Environment - Polygenic Risk Scores
• Comparing whole genome polygenic risk scores to polygenic risk set scores is difficult as the 'ideal' significance threshold for a whole genome polygenic risk score is unlikely to be the same for a polygenic risk set score, and will likely change depending on which set you test.
• SAVE-PRS allows for "real-time" comparison of set scores to whole genome polygenic risk scores in a shiny app, to allow comparison across all significance thresholds and allow better interpretation of all results.

Current datasets tested in PRS analysis:

PGC2 (SCZ)

PGC2minusCLOZUK (SCZ2minusCLOZUK)

Location on rocks: /home/SHARED/PGC/daner_PGC_SCZ52_0513a.resultfiles_PGC_SCZ52_0513.sh2_noclo.txt
Summary stats

Sweden+SCZ1

Location: https://www.med.unc.edu/pgc/results-and-downloads
Summary stats

SCZ1

Location: https://www.med.unc.edu/pgc/results-and-downloads
Summary stats
WARNING build is UCSC hg18, will likely need to convert to hg19.

CLOZUK

CLOZUK_no_PGC

Location on raven: /neurocluster/databank/CLOZUK/GWAS/BGE/*CLOZUK_GWAS_BGE*.tar.gz
Best guess genotype data

CLOZUK_Meta-analysis with PGC

Location on raven: /neurocluster/databank/CLOZUK/GWAS/SUMSTATS/CLOZUK_PGC2noclo.wCHRX.w1000Gfrq.METAL.assoc.dosage.gz
Summary stats

ALSPAC

Best guess genotype data

Biobank wv 1

Best guess genotype data

Edit the parallel commands to reduce the number of cores that are used to process in SurPRSe which will reduce the errors produced by this command.