This document contains notes regarding the UK Biobank PRS Release described under Category 300 in the UK Biobank Data Showcase (https://biobank.ndph.ox.ac.uk/ukb/label.cgi?id=300).
Version 1 of the UK Biobank PRS Release was released in May 2022.
Version 2 of the UK Biobank PRS Release was released in May 2024, and incorporates feedback from users, and new data.
The polygenic risk scores (PRSs) are described in Thompson et al. 2022 (https://doi.org/10.1101/2022.06.16.22276246), and the performance evaluation of the UK Biobank PRS Release v2 is described in Thompson et al. 2024 (https://doi.org/10.1371/journal.pone.0307270).
For any questions or comments regarding these files, please contact Genomics plc at research@genomicsplc.com .
- Subgroup on whom Standard PRS scores are calculated. Following new UKB withdrawals, the number of UKB individuals on whom Standard PRS scores are calculated has reduced by 46. N=486,176.
- Subgroup on whom Enhanced PRS scores are calculated. Following feedback from users, the subgroup of UK Biobank for whom Enhanced PRS scores are calculated has been expanded to include all individuals in UK Biobank who are not members of White British Unrelated subgroup (the latter was used to generate GWAS training data for the Enhanced PRS set - see Thompson et al. 2024. N=148,936.
- Testing subgroup for PRS performance evaluations. The Testing Subgroup has been altered to satisfy the following criteria: (a) not part of the White British Unrelated subgroup; (b) not closely related to any individual in the White British Unrelated subgroup (no relatives of 2nd degree relatedness or above with the WBU sample were included, KING kinship coefficient threshold = 2-3.5 = 0.0884). N=119,879 (EUR=97,608, SAS=9,542, AFR=9,476, EAS=2,864, AMR=389). The v2 Testing Subgroup differs from the v1 Testing Subgroup, both by individuals added and by individuals taken away. Comparisons of performance metrics between the two Testing subgroups indicate similar values and no biases in reported metrics (see “v1 Dataset release notes, issues and clarifications” below).
- Enhanced PRS for Crohn’s disease and ulcerative colitis. For reasons explained in Thompson et al 2022 (Supplementary Materials), the Enhanced PRS score in the Enhanced PRS subgroup for Crohn’s disease and ulcerative colitis has been set equal to the Standard PRS score for these two diseases.
- Updated Standard and Enhanced PRSs - new data. Following the ingestion of new external summary statistic GWAS data, updated or new Standard and Enhanced PRS scores have now been created for the following quantitative traits: glycated haemoglobin, resting heart rate, low density lipoprotein cholesterol, high density lipoprotein cholesterol, total triglycerides, and total cholesterol. Of these, the Standard PRSs for resting heart rate, total triglycerides, and total cholesterol are new (Version 1 did not include Standard PRSs for these traits).
- Updated Standard and Enhanced PRSs - new centering. The prostate cancer PRS (Standard version) had suboptimal centering properties in Version 1. This issue has been fixed in Version 2.
-
All PRS in the release are meant to be centered. We note however that the prostate cancer PRS (Standard version) has suboptimal centering properties. This issue will be fixed in the next release.
-
Both Crohn's disease and ulcerative colitis are primarily based on a relatively sparse ImmunoChip array. This sparsity complicated the meta-analysis process, leading to less effective Enhanced PRS. For this reason, the Enhanced version of the PRS is not provided for either of these two diseases. Please use the Standard versions in lieu of the Enhanced versions.
-
Motivated by the evaluation requirements for the Enhanced PRS set, which uses the White British Unrelated (WBU) group for training, we defined our Testing group to exclude individuals related to the WBU group. This process resulted in a Testing group of 104,621 individuals. However, we noted that this process was not perfectly applied. Out of the 104,621 individuals in Testing, 28,177 are related above 3rd degree level (kinship > 2^-4.5) and 12,229 are related above 2nd degree level (kinship > 2^-3.5) to individuals in WBU. We compared PRS performance in EUR Testing individuals with and without relatedness to WBU and this did not measurably impact the evaluation (see figures below). This issue will be fixed in the next release. Meanwhile, users keen to investigate further can identify individuals related to the WBU group using the relatedness file described in https://biobank.ndph.ox.ac.uk/ukb/refer.cgi?id=531.
This Return-of-Results package relates to the following paper: Thompson et al. 2022. UK Biobank release and systematic evaluation of optimised polygenic risk scores for 53 diseases and quantitative traits. https://www.medrxiv.org/content/10.1101/2022.06.16.22276246v1
The results data were generated under UK Biobank project number 9659 (under its extension to therapeutic applications).
Our paper describes the UK Biobank PRS Release: a set of polygenic scores for 53 disease and quantitative traits, together with associated phenotypic and genetic information (see below for list of traits).
A separate tool (UKB-PRET) is provided to facilitate evaluation in a testing subset of the UK Biobank. See https://github.com/Genomicsplc/ukb-pret
The polygenic risk scores (PRSs) are described in Thompson et al. 2022 (https://www.medrxiv.org/content/10.1101/2022.06.16.22276246v1). Data supporting these scores were obtained either entirely from external GWAS data (the Standard PRS set) or from a combination of external and internal UK Biobank data (the Enhanced PRS set). This work was conducted by Genomics PLC under UK Biobank project 9659."
The Release is packaged as a series of data files, with fields as described below.
PRS genetic principal components – field 26201
In UKB WBU PRS testing set – field 26200
PRS fields:
One for each trait and PRS type (Category 302 - Enhanced (also known as UKB-WBU) and Category 301 - Standard (also known as UKB-Free)))
PRS genetic principal components: the first four PC axes obtained from the common genotype data in the 1000 genomes reference dataset using standard methods (Price et al, 2006) are provided in the four array indices in field 26201.
In UKB WBU PRS testing set: indicator of participants in the UK Biobank PRS Release testing set
prs: polygenic score for the trait for every participant in the UK Biobank PRS Release Testing subgroup, as described in the paper
prs: polygenic score for the trait for all UK Biobank participants, as described in the paper
The following is a list of traits and their phenotype codes (as used in file naming). See also src/ukb_pret/data/traits.yaml
DISEASE TRAITS
Age-related macular degeneration AMD
Alzheimer's disease AD
Asthma AST
Atrial fibrillation AF
Bipolar disorder BD
Bowel cancer CRC
Breast cancer BC
Cardiovascular disease CVD
Coeliac disease CED
Coronary artery disease CAD
Crohn's disease CD
Epithelial ovarian cancer EOC
Hypertension HT
Ischaemic stroke ISS
Melanoma MEL
Multiple sclerosis MS
Osteoporosis OP
Prostate cancer PC
Parkinson's disease PD
Primary open angle glaucoma POAG
Psoriasis PSO
Rheumatoid arthritis RA
Schizophrenia SCZ
Systemic lupus erythematosus SLE
Type 1 diabetes T1D
Type 2 diabetes T2D
Ulcerative colitis UC
Venous thromboembolic disease VTE
QUANTITATIVE TRAITS
Age at menopause AAM
Apolipoprotein A1 APOEA
Apolipoprotein B APOEB
Body mass index BMI
Calcium CAL
Docosahexaenoic acid DOA
Estimated bone mineral density T-score EBMDT
Estimated glomerular filtration rate (creatinine based) EGCR
Estimated glomerular filtration rate (cystatin based) EGCY
Glycated haemoglobin (excluding participants with diabetes) HBA1C_DF
High density lipoprotein cholesterol HDL
Height HEIGHT
Intraocular pressure IOP
Low density lipoprotein cholesterol (statins-free individuals) LDL_SF
Omega-6 fatty acids OSFA
Omega-3 fatty acids OTFA
Phosphatidylcholines PDCL
Phosphoglycerides PHG
Polyunsaturated fatty acids PFA
Resting heart rate RHR
Remnant cholesterol (Non-HDL, Non-LDL cholesterol) RMNC
Sphingomyelins SGM
Total cholesterol TCH
Total fatty acids TFA
Total triglycerides TTG