This is a summary of the data processing and plots created in the DOSCHEDA shiny app.

Number of Channels | Number of Replicates | Input Type | Model fitted ---|---|---|--- r params$object@parameters$chans|r params$object@parameters$chans | r | r ifelse(params$object@parameters$modelType == 'sigmoid','Sigmoidal','Linear')

   # reptest <- ifelse(params$reps > 1,TRUE,FALSE)
ex<- params$object
ob_params <- getParameters(ex)
if(ob_params$reps == 1){
  reptest <- FALSE

  reptest <- TRUE

Figure 1: Box plot of Data Columns



Here we should observe the means centered at approximately zero across all channels and replicates.

Figure 2: Ranked plot of proteins FC and Density plot of Data Columns



These plots show the density distribution of each channel and the distribution of the ranked proteins. The Density plots should have a bell shaped distributions and be approximately centred at zero.

Figure 3: Mean vs standard deviation plot



The ranked row means versus the standard deviations for checking any dependence of the variance on the mean. The red line is a running median estimator with window-width at 10%. If this red line is approximately horizontal, this indicates no variance-mean dependence.

Figure 5: Pearson Correlation between Conditions



The Pearson correlations (r) between all the different channels are displayed. None of the channels are expected to be anti-correlated (r < 0). The QC in DOSCHEDA will highlight whether there are anti-correlated pairs of channels.

Figure 6: Principal-Component Analysis of Data columns



Plot of two highest principal components. Replicates should (roughly) cluster together. This plot highlights if any data points are vastly 'out of place' given the experimental design.

Figure 7: Replicate vs Replicate plots for each Channel.

ob_param <- getParameters(ex)
for (i in 1:ob_param$reps) {
  for  (j in 1:ob_param$reps){
    if(i >= j ){
      for (k in 0:(ob_param$chans - 1)){
          replicatePlot(ex,conc = k,repIndex1 = i,repIndex2 = j)


Scatterplots between replicates to identify targets (drug competed proteins) and potential off-targets. Points that have high fold change in both replicates and are close to the red x = y line are considered to be targets.

if(ob_param$modelType == 'sigmoid'){
  show_text <- TRUE
} else{

  show_text <- FALSE

```{asis, echo=!show_text}

Linear Model Plots

The following plots are relevant if a linear model has been fitted to the data.

Figure 8: Distribution of p-values of model coefficients


```{asis, echo=!show_text} Description:

The p-value distributions for each of the model coefficients are expected to not be uniform. The QC in DOSCHEDA will highlight whether a coeffcient does not contail any significant p-values.

Figure 9: Volcano plots of model coefficients

    for (i in c('slope','intercept','quadratic')){
      volcanoPlot(ex,coefficient = i)

```{asis, echo=show_text}

Sigmoidal Fit Plots

The following plots are relevant when a sigmoidal model is fitted to the data. They show the top protein profiles for each of the model parameters.

```{asis, echo=show_text}
**Figure 8: Plot of the largest differences between the proteins from the lowest and highest concentrations (over 30%)**
plot(ex,sigmoidCoef = 'difference')

```{asis, echo=show_text} Description:

Plot of the proteins profiles whose difference between the top and bottom parameters of the sigmoidal model are greater than 30%.

```{asis, echo=show_text}
**Figure 9: The top proteins with significant Slope Value**
plot(ex,sigmoidCoef = 'slope')

```{asis, echo=show_text} Description:

The top 15 protein profiles with a significant slope parameter of the sigmoidal fit.

```{asis, echo=show_text}
**Figure 10: The top proteins with significant RB50 values**
 plot(ex,sigmoidCoef = 'rb50')

```{asis, echo=show_text} Description:

The top 15 protein profiles with significant RB50 parameter of the sigmoidal fit.


Try the Doscheda package in your browser

Any scripts or data that you put into this service are public.

Doscheda documentation built on Nov. 8, 2020, 5:37 p.m.