Components of genetic associations across 2,138 phenotypes in the UK Biobank highlight adipocyte biology

Tanigawa et al., Nature Communications 2019

Motivated by the polygenicity and pleiotropy of genetic associations in complex traits, we developed a method called Decomposition of Genetic Associations (DeGAs) that operates on a series of genome-wide summary statistics across multiple phenotypes. Please check our publication for more information. Yosuke also summarized highlights of the study here.

Decomposition - Methods overview

Y.T. developed the DeGAs web application. M.A.R supervised computational aspects of the study.

Video Tutorial for DeGAs App

Yosuke explains the DeGAs app and teaches viewers how to interpret scores and plots using examples. The coding variants is used as an example.

The analysis codes and datasets used in the DeGAs paper

The analysis scripts used in the study are available on our GitHub repository.

The 3 datasets analyzed in the study are available on figshare.

Tanigawa et al. 2019, Nature Communications datasets

We provide the interactive DeGAs app for the three datasets analyzed in the paper:

Other available datasets in the DeGAs app

Additionally, following datasets with different parameters are also available in our DeGAs app:

Naming convention

The name of the datasets corresponds to the following conditions:

  • all, coding, non-coding, and PTVs: This indicates the type of variants included in the analysis. PTVs = Protein Truncating Variants, coding = coding variants, non-coding = non-coding variants (does NOT include coding variants), and all = coding + non-coding.
  • nonMHC: By default, all the variants on autosomes are included in the analysis. nonMHC dataset has variants outside of major histocompatibility complex region.
  • z: The elements of summary statistic matrix is Z-score
  • non-center: By default, the left/right-singular vector from truncated-singular value decomposition is centered around zero. non-center datasets skip this procedure.
  • p[some value]: The default p-value threshold is 0.001. If there is a suffix like 'p0.01', that means the p-value threshold is adjusted to the specified value.