Figure 4

Hemolysis and tobacco use are factors that broadly impact plasma miRNA stability. (A) Measurement bias (mean deviation in log₂ expression away from the predicted longitudinal trend) in the intraindividual expression levels for each of the 134 miRNAs that passed filtering, as associated with hemolysis of the plasma, self-reported measures of tobacco use, fasting status, higher stress level, and higher quality of last sleep. (B) Kernel density representation of the count distribution of the miRNAs from panel (A) relative to the average change in log₂ expression, by associated factor. (C) Test–retest variance moderation (mean deviation in log₂ SD magnitude away from the sample-average value) for each of the 134 miRNAs that passed filtering, as associated with hemolysis of the plasma, participant age per decade of life, higher average quality of sleep over the sampling window, higher average stress, higher probability of fasting, participant sex, and regular tobacco use. (D) Kernel density representation of the count distribution of the miRNAs from panel (C) relative to the average change in log₂ test–retest SD, by associated factor. Analyses for each factor included adjustments for all the other factors (e.g., hemolysis results for measurement bias are adjusted for tobacco, fasting, stress, and sleep).