Tanigawa, Yosuke Sinnott-Armstrong, Nasa Rivas, Manuel The multi-PRS weights computed with the 35 lab biomarkers described in 'Genetics of 35 blood and urine biomarkers in the UK Biobank' <p>The dataset contains the multi-PRS weights computed with the 35 biomarker traits described in the following preprint:<br>N. Sinnott-Armstrong*, Y. Tanigawa*, et al, Genetics of 38 blood and urine biomarkers in the UK Biobank. bioRxiv, 660506 (2019). doi:10.1101/660506<br><br>Note that we are preparing a revised version of the manuscript and this dataset contains 35 (instead of 38) biomarker phenotypes.<br><br>The list of disease endpoints included in this dataset is: angina, alcoholic cirrhosis, gallstones, hypertension, cholecystitis, kidney failure, heart failure, myocardial infarction, gout, and type 2 diabetes (T2D).<br><br>We provide weights of the 23 polygenic risk scores characterized by multi-PRS. The list of models are summarized in list_of_multi-PRS-models.tsv. This index file the following columns:</p> <ul><li>Filename: the filename of polygenic risk score weights in this dataset.</li><li>Trait: the disease outcome.</li><li>Covariate_adjustment: a binary variable indicating whether the multi-PRS model is trained with covariate (age, sex, and PC1-10) adjustment.</li><li>Family_history_adjustment: a binary variable indicating whether the multi-PRS model is trained with family history.</li><li>Note: Additional information when relevant.</li></ul> <p>For the PRS models listed with "Covariate_adjustment == TRUE", we fit multi-PRS regression model adjusted by age, sex, and 10 principal components where as the ones with "Covariate_adjustment == FALSE" we did not use those covariates.<br><br>For T2D, we have two sets of models: (1) models trained for Eastwood et al. T2D cases in UK Biobank and (2) models trained for Eastwood et al. T2D cases in UK Biobank vs. filtered controls with HbA1c < 39.<br><br>For myocardial infarction, we provide a model with family history adjustment, <code>weights_familyhistory.HC326.tsv.gz</code>. This model is trained with covariates (age, sex, and 10 principal components) and family history of heart disease as covariates.<br><br>Please read our manuscript for more details.</p> <p>For each PRS model listed in <code>list_of_multi-PRS-models.tsv</code>, we provide a compressed tab-delimited files, which contain the multi-PRS weights. The files have the following columns:</p> <ul><li>CHROM: the chromosome</li><li>POS: the position</li><li>ID: the variant identifier</li><li>REF: the reference allele</li><li>ALT: the alternate allele</li><li>A1: the risk allele</li><li>weights.: the coefficients (weights) of the PRS</li></ul> <p>Note that we used GRCh37/hg19 genome reference in the analysis and the BETA is always reported for the alternate allele.<br><br>The multi-PRS weights files are compressed with <code>gzip</code>. One should be able to read those files with the standard <code>gzip</code>/<code>zcat</code>.</p> Polygenic risk score (PRS);UK Biobank;Angina;alcoholic cirrhosis;Gallstones;Hypertension;Cholecystitis;Kidney failure;Heart failure;Myocardial Infarction;Gout;T2D;Type 2 Diabetes 2020-06-19
    https://nih.figshare.com/articles/dataset/The_multi-PRS_weights_computed_with_the_35_lab_biomarkers_described_in_Genetics_of_35_blood_and_urine_biomarkers_in_the_UK_Biobank_/12355424
10.35092/yhjc.12355424.v1