We develop PGSFusion to perform polygenic score (PGS) construction and epidemiolgical application. PGSFusion accepted many different formats of summary statistics and implements a wide range of PGS methods, including single-trait, multiple-trait, annotation-based, and cross-ancestry methods. PGSFusion provided further epidemiological analysis: i) prediction performance; and ii) joint analysis.
The user can use the server to visit the website: http://www.pgsfusion.net/#/
- We select 138 quantitative traits (i.e. 61 Anthropometric traits, 65 Blood and Unrine Test traits and 12 Other traits) and 63 binary traits (i.e. 14 Cancers, 44 Non-cancer Diseases, and 5 Other traits).
- We use three variables at baseline to assess the SES of each participant at individual level: including family income level (data field: p738_i0), education qualification (data field: p6138_i0), employment status (data field: p6142_i0).
- We include information on four healthy lifestyle factors collected at baseline, including “no current smoking”, “regular physical activity”, “healthy diet”, and “no alcohol consumption”.
- We used UKBB data to build the reference panel, validation set and test set. For EUR reference panels, we used 2,000 EUR individuals in UKBB. For AFR and EAS, we used corresponding individuals in 1000 Genome Project. For EUR test set, we selected 50,000 individuals in UKBB. For AFR and EAS, we selected all the corresponding samples in UKBB.
- We make refenence panel for different methods using 2,000 EUR individuals from UKBB.
- We make the reference panel for EUR-AFR and EUR-EAS for SDPRX.
- We integrated 17 PGS methods, including 11 single-trait, 1 multiple-trait, 2 annotation-based and 3 cross-ancestry based methods.
- We constructed the reference panel for each method.
- We output the linkage for the selected methods
- We provide two further applications: i) performance evaluation; ii) joint analysis.
- Add the number of quantitative and binary traits
- Update DBSLMM (v1.0)
- Use 11 strata variables
- Add SBayesRC method
- Add the comparison of prediction performance for selected methods
- Use 2,000 EUR individuals to build EUR reference panel
- Update the PGSFusion in formating summary statistics and performing epidemiological applications
- Build the PGSFusion to streamline the PGS construction
If you used our server, please cite the our paper:
Yang S#*, Ye X#, Ji X#, Li Z, Tian M, Huang P*, Cao C*. PGSFusion streamlines polygenic score construction and epidemiological applications in biobank-scale cohorts. bioRxiv.