Fig. 5
From: Multiple freeze-thaw cycles lead to a loss of consistency in poly(A)-enriched RNA sequencing
Higher Sample Quality has Lower Discordance at Each Subset Size. Linear regression estimates of discordance predicted from effect size and subset quality for each subset size. High sample quality (blue) is compared to low sample quality (red). ΔD values represent the change in discordance per unit increase in effect size. a The predicted discordance with respect to the mean-variance standardized effect at a subset size of 14; sample quality is assessed by freeze-thaw. The disparity (Δm) between the change in discordance per unit increase in effect size for high (DH) and low (DH) quality subsets is also displayed. Summary of results for each subset size (x-axis) for sample quality represented by either (b) freeze-thaw or (c) RIN. Significant estimates (Wald test, Benjamini-Hochberg FDR correction, q ≤ 0.1) are marked by a circle and insignificant estimates are marked by a cross. For (b) freeze-thaw, Δm corresponding to panel (a) is also displayed