The above table provides results of partitioned LD score regression (LDSR) for GWAS of 4178 UK Biobank phenotypes. You can read more about this analysis on the methods page and find links to more information about the underlying GWAS on the Neale Lab blog.

Additional fields can be viewed using the “Select Columns” button at the top left (columns described below). Use the search box at the top right to filter the table.

You can also download the complete results file.

Search Suggestions

Use the search box on the top right of the table to filter the table.

Find phenotypes by name

  • Description, e.g. height or "Time spent"
  • ICD code, e.g. I25
  • ID, e.g. 2734, i.e. the phenotype ID matching the UK Biobank showcase

Focus on a certain results category

Searches include all non-numeric fields, so you filter on flags in multiple columns. For example, try

  • z7 finngen for highly significant \(h^2_g\) results for ICD10-based endpoints defined by FinnGen
  • female medication for medication codes where the primary GWAS is in females only

Construct complex searches

Your search can include JavaScript-style regular expressions. These are especially useful for combining keyword options or using e.g. the word boundary \b to delineate strings that might appear elsewhere. For example, try

  • \bz[47] \bmedium \bmale for medium confidence results with \(z>4\) or \(z>7\) for \(h^2_g\) and where the primary GWAS is in males
  • (beer|wine|alcohol) for all phenotypes containing the terms “beer”, “wine”, or “alcohol”
  • "blood pressure" ^((?!(medication)).)*$ for all phenotypes containing “blood pressure” but not “medication”

Table Columns

The above browser includes the following fields:

Name Description
ID (hidden) either the UK Biobank phenotype code (e.g. matching the UKB Showcase), an ICD10 code, or a FinnGen endpoint code
Phenotype Name/description of the phenotype. Clink to link to a more detailed set of LDSR results for the phenotype.
h2 the SNP heritability estimate from partitioned LD score regression. Value is given on the liability scale for binary phenotypes, assuming that the population prevalence matches the prevalence in the UK Biobank analysis set.
h2 se (hidden) standard error of the SNP heritability estimate.
h2 p p-value for the test that the SNP hertability is greater than 0.
h2 sig? whether the SNP heritability estimate is statisitical significant at different thresholds. Possible values are z7 (z > 7, highly significant), z4 (z > 4, significant after correction for multiple testing), nominal (p < .05), nonsig (p > .05), or NA (low confidence results). See methods for more details.
Confidence confidence rating for the SNP hertiability estimate. One of low, medium, high, or none. See methods for more details.
Notes (hidden) explanation of confidence ratings lower than “high”, as a list of flags delimited with semi-colons. Possible flags are
Int. the intercept term from partitioned LD score regression. Values > 1 may indicate population stratification, confounding, or some other model misspecification.
Int. se (hidden) standard error of the estimated LD score regression intercept.
Int. p p-value for the LD score regression intercept testing whether the intercept is greater than 1.
Lambda (hidden) (Median observed \(\chi^2\))/(Median of 1 degree of freedom \(chi^2\) distribution). Genomic control measure of inflation of GWAS results for the SNPs used in the LDSR analysis. Increases with sample size.
Ratio (hidden) (Estimated LDSR intercept -1)/(Mean observed \(\chi^2\) -1). Measure of the proportion of inflation of GWAS results that is not explained by the relationship with LD modelled by LDSR. May reflect remaining stratification, confounding, structure, or model misspecification.
Neff The effective sample size in the UKB GWAS analysis set. For binary variables is 4/(1/\(N_{case}\))+(1/\(N_{control}\)), otherwise is the sample size. More info on the GWAS analysis set is available on the Neale Lab blog.
Var Type (hidden) Measurement type for the phenotype. One of binary, ordinal, continuous_irnt (i.e. inverse rank-normalized), or continuous_raw.
Prev. (hidden) prevalence for binary phenotypes. Is blank if the variable isn’t binary.
Source the GWAS batch the phenotype belonged to. One of: PHESANT, ICD10, FinnGen, or biomarkers.
Sex (hidden) the sex strata used as the primary GWAS for this phenotype, e.g. to limit choice of controls for sex-specific case/control phenotypes. One of: both, female, or male.

Hidden fields can be accessed using the “Select Columns” botton on the top left of the browser table. Search includes all fields even when hidden.

Click on a phenotype for more detailed results from the partitioned LD score regression analysis.