Fig. 6

An example of an R/LinkedCharts app (C, D) for a simple exploratory analysis and the code to generate it in comparison with static plots (A, B) produced for the same purpose. The heatmaps (A, C) show Spearman correlation of gene expression for all samples from Conway et al. [15]. Here, we can see, inter alia, two outlier samples in the heatmap’s bottom-right corner and some more or less pronounced clusters of samples with similar gene expression levels. The scatter plots (B, D) show expression values for two samples plotted against each other. Browsing through several such plots can help the researcher get a feeling of the data and explore unexpected patterns like the outliers just mentioned. The code is split into two pieces, where the upper one is responsible for generating the plots and the lower part shows the code to update them to show a specific sample pair. For static plots, one has to execute the same lines of code for any pair of samples, while for R/LinkedCharts the provided code should be added to the list of arguments for the heatmap. After that, switching between pairs of samples can be done simply by clicking on the corresponding cell of the heatmap. The static heatmap (A) was generated with the “pheatmap” package [30]; scatter plot (B) was made with a base R function. The live version of the app can be found in the supplement. For simplicity, gene expression for all the samples is subset to 8000 randomly selected genes