Skip to main content
Fig. 1 | Molecular Medicine

Fig. 1

From: Gene expression variability in long-term survivors of childhood cancer and cancer-free controls in response to ionizing irradiation

Fig. 1

Comparison of methods for quantification of expression variability and overview of distribution and subsequent variability classification: A Comparison of different metrics that are typically used for the characterization of expressional variability. Scatter plots showing the association of the coefficient of variation (CV, standard deviation/mean), standardized median absolute deviation (d, median absolute deviation (MAD)/median), and expression variability (EV, computed as MADobserved-MADexpected) on the respective y-axes with the median expression on the x-axis. Kendall’s correlation coefficient (τ) for the correlation of each variability metric with the median gene expression is provided per facet. The yellow stripes on the left show the value distribution of each variability metric. The colours of the dots represent the donor groups (N0 = fibroblasts of cancer-free controls, N1 = fibroblasts of childhood cancer survivors without a second primary neoplasm, N2 +  = fibroblasts of childhood cancer survivors with at least one second primary neoplasm) Below, the distribution of the median gene expression is presented as box- and density plots. B Density plots showing the distribution of the EV (truncated using the smallest Q01 (N0 after LDIR) EV = − 427.9 to the largest Q99 (N0 after 2 Gray) EV = 1204.8 for reasons of legibility), also stratified by donor group, and radiation dose. Dashed lines imply the ranges employed for the classification into hypo-, non-, and hyper-variable, whereas genes with an EV lower than the left dashed lines were classified as hypo-variable and genes with an EV higher than the right dashed line were conversely classified as hyper-variable. Genes with an EV between the dashed lines were classified as non-variable. C Bar charts showing the number of genes that were classified as hypo-, non-, and hyper-variable, using the whole data set and the number of genes with stable classification confirmed by cross-validation. All data were stratified by exposed radiation dose (0, 0.05, and 2 Gray)

Back to article page