Fig. 5

LinkedCharts can be used to “walk backwards” through an analysis pipeline. This is illustrated here using a drug screening experiment [23, 24] as an example. For an interactive version, see Interactive Supplementary Fig. 4. The blue arrows show the direction of a typical analysis pipeline used in drug screening experiments. We start with reading intensity values from plates with different cell lines cultured in the presence of studied drugs (A). These values are then normalized and turned into a fraction of the cells that remained viable. A sigmoid curve is fitted to the obtained viability values at different drug concentrations, and the area under the fitted curve yields a single score for each drug (B). Different drugs’ scores are compared to each other across all the tested cell lines (C). A drug-drug correlation heatmap is then produced to identify clusters of similar drugs (D). The red arrows illustrate the direction of interactive data exploration: We start by showing the summary heatmap plot (D). Suppose the researcher is interested in a particular drug combination or a cluster of drugs. In that case, he or she can examine the corresponding drug scores simply by clicking on the heatmap cell (D) to see the underlying correlation plots (C). Similarly, one can click in a point in (C) to examine the individual viability values at the tested concentrations and check the sigmoid fit (B). And finally, if needed, it is possible to take one more step back and to look at the raw read-outs to inspect them for the presence of any artifacts (A)