Fig. 2
From: Machine learning identifies a compact gene set for monitoring the circadian clock in human blood
Properties of the predictor trained on all control samples from the three datasets (sumabsv = 2, nSPC = 2). a Expression of the two SPCs vs. circadian time. Each point is a sample. Black curves correspond to periodic smoothing splines fit by ZeitZeiger. The signal-to-noise ratios of the two SPCs are 3.06 and 2.01, respectively (compare to Additional file 1: Figure S1). b Genes and coefficients for the two SPCs. Genes are sorted by their respective coefficients. The expression of a given SPC in a given sample corresponds to the dot product of the coefficients for that SPC and the gene expression for that sample