hirotaka-i / amp-pd-v3-pc-prs

AMP-PD version 3: PC and PRS calculation

Please see main.ipynb for the actual processing of data.

We do not share the individual level data here. Please visit https://amp-pd.org/ for the data access.

PC calculation and population assignment

AMP-PD version 3 wgs genetics data of 10,418 samples were merged with the 1000 Genomes Project Phase 3 reference panel. Using the mean +/- 6SD threshold of PC1-5, the ancestry {EUR, EAS, AFR, OTHER} was assigned.

Ancestry assignment results:

• EUR 10119
• EAS 32
• AFR 111
• OTHER 156

PRS calculation

The polygenic risk scores (PRS) were calculated based on the PD GWAS summary statistics from Nalls et al. 2019 (PMID: 31701892). The weight file was downloaded from https://github.com/neurogenetics/genetic-risk-score. The file contains the weights of 90 risk SNPs but two SNPs were not available in the AMP-PD genetics data. We generated various versions of the PRSs and in some versions, the missing SNPs-rs11578699 was substituted by rs11577197 (perfect LD) and rs3742785 was substituted by rs10134885 (R2=0.989). The PRSs were normalized (N[0, 1]) across the AMP-PD for the ease of use.

PRS variations:

• PRS88 -> Full (but missing 2 SNPs)
• PRS83 -> PRS88 Without GBA, LRRK2 loci
• PRSp90 -> Used proxy above
• PRSp85 -> Used proxy above without GBA, LRRK2 loci
• PRSp87 -> Used proxy above without GBA locus
• PRSp88 -> Used proxy above without LRRK2 locus

Pleaes note that the PRS weights were derived from the European GWAS results. Therefore, the PRSs may be not easily generalizable to non-European population.