Fig. 4: RGStraP was also tested on samples from the Geuvadis cohort, which includes 5 populations of European and African ancestries.
From: Direct inference and control of genetic population structure from RNA sequencing data
a Canonical correlation analysis between ten RG-PCs and ten array PCs of the Geuvadis samples showed significant (Wilks’ Lambda, p-value < 0.05) correlations for the first 4 canonical variates (CVs) between the two sets. The first 3 CVs from 10 RG-PCs strongly captured the genetic information from array PCs (Rc1 = 0.994, Rc2 = 0.942, Rc3 = 0.752). The cumulative proportion of shared variance between the three sets reached 0.998 from just the 3 CVs, 0.903 of them represented by CV1. b PCA plots of the Nepal and Geuvadis samples showing comparable population structure between the array-based PCs and RG-PCs, separating the ancestry groups (European, African, and South Asian) in the main PCs.
