Fig 1.
Comparison of standard deviations for geNorm and LEMming per gene in DS1 and DS2.
The x-axis shows the difference between standard deviation (sd) of raw values and sd of geNorm processed data per gene. Accordingly the y-axis shows the difference of sd of raw values to LEMming processed data. If points are in the positive quadrant and above the dotted line, sd of LEMming processed data is smaller compared to sd geNorm and raw data.
Fig 2.
Example boxplot for Foxo1 in DS2.
Left: Log2-fold differential expression of the gene Foxo1 in DS2. Conditions are 0h, 24h, 48h and 72 h. Right: Variance-mean plot showing the mean over the standard deviation per condition for Foxo1.
Fig 3.
Raw data of common reference genes in data set 3.
Boxplots of the untreated conditions are black, boxplots of treatment conditions (dedicated in S4 File) that are not significant differentially expressed compared to untreated are blue and boxplots of treatments with significant differentially expressed measurements are red. Significance was calculated by an unpaired t-test with unequal variances and Bonferroni corrected significance level α = 0.05/48. Outliers are marked by red circles.
Fig 4.
Technical and biological variances over mean value per gene.
(a) Standard deviation of Ct value of each gene and well over the mean Ct value for data set 1 (DS1). Each gene is measured six times per biological replicate (3× cDNA and 2× PCR per cDNA). The regression line shows that the standard deviation increases with the Ct values (lower mRNA content). (b) Standard deviation of biological replicates over the mean Ct value for DS1. The mean of all six technical replicates is computed per biological replicate and gene. The standard deviation of these means is computed with 4 biological replicates for each gene. The biological variance is higher under treated conditions compared to the control conditions.
Fig 5.
Contribution of biological variance, cDNA conversion error and qPCR error to the overall variance for data set 1.
Proportions of variance contribution are estimated from raw data (a) and from residuals of LEMming (b). Blue boxplots are measurements of the control group, red boxplots are measurements of WY14,643 treated cells.
Fig 6.
Proportional contribution of different effects to the variance of a gene in data set 2 (DS2).
Black—raw data, Green—LEMming processed data. (a) Proportion of sum of squares associated to the effects time, primer pipetting, biological variance, cDNA conversion, qPCR error and sample pipetting error (SPE) resulting from a ANOVA for each gene. (b) Proportion of sum of squares of LEMming processed to raw data without the effect of time (treatment effect). The median is 16.9%, which means that LEMming excludes systematic effects that are responsible for 83.1% of variance of the median gene in this experiment.
Fig 7.
Distribution of raw data and residuals in reference genes in DS3.
Density plot of raw data (a) and residuals of LEM-method (b) of reference genes in DS3. Blue: kernel density estimation of raw data/residuals. Red: estimated Student-t distribution. (c) Quantile-Quantile plot with quantiles of estimated Student-t distribution versus quantiles of residuals.