Skip to main content
. 2021 Jul 23;15:47. doi: 10.1186/s40246-021-00347-y

Table 1.

Principal component variance analysis (PCVA) decomposition of sources of variance

Gene expression PSI
(108,091 exon bins)
PSI
(7001 exon bins)
Principal component
  PC1 28.9% 7.6% 9.3%
  PC2 25.1% 5.5% 6.9%
  PC3 5.2% 4.1% 4.1%
  PC4 4.4% 3.0% 3.2%
  PC5 3.5% 2.6% 2.9%
  PC6 2.3% 2.4% 2.7%
  PC7 2.0% 2.3% 2.3%
  PC8 1.6% 1.8% 2.0%
  PC9 1.6% 1.7% 1.9%
  PC10 1.3% 1.7% 1.6%
  SUM of PC1–PC10 75.9% 33.0% 27.9%
Variance component
  Ancestry 2.3% 5.8% 5.3%
  Location 0.2% 0.7% 0.5%
  Disease 40.3% 20.7% 27.9%
  Location*disease 0.8% 0.8% 1.0%

The table shows the percent variance in gene expression or splicing (PSI at two inclusion thresholds) explained by the first ten PCs and their sum, as well as the weighted contribution of each variance component term (ancestry, disease, location, and the interaction of disease and location) to these 10 PCs. Gene expression is for 18,929 genes, and number of PSI bins is before and after the final two stages of filtering (n = 119 samples for all)