The goal of using CMap is to find connections to your query signature that enable you to generate hypotheses for further experiments. In general, we first look closely at query results with connectivity scores >+95 and <-95. But if the CMap database contains many signatures related to the signature of your query, you may recover a number of connections with scores greater than 95. These top connections can be of varied MoA or gene function and you'll need to filter the list for those that seem most relevant for your research. As it can be challenging to sort through a large list of high scores, we've developed some tools and methods for exploring the top hits to help you determine which perturbagens to validate in further studies.
As an example, the plant flavonoid apigenin is known to have both anti-inflammatory and anti-proliferative activities. To see if CMap can help pinpoint the MoAs of apigenin, we can take a look at its connections using the Touchstone app, as apigenin is in the Touchstone dataset. The figure below shows a heatmap view of the top compound connections to apigenin, sorted from highest- to lowest-ranked connections by the "summary" column, which represents connectivity scores that have been summarized across cell lines (connections in the individual cell lines are shown as well).
You can see the scores by mousing over each cell of the heatmap. While it's apparent that over 40 compounds have connectivity scores of 98 or greater, scanning down the descriptions reveals many strong connections to CDK (cyclin-dependent kinase) inhibitors, suggesting a hypothesis that apigenin has CDK inhibitor activity, which could explain its effect on cell proliferation.
To further explore this hypothesis, filter the list by perturbational class (PCL) using the option the Quick Tools window of the heatmap. A PCL is a group of compound or genetic perturbagens that have the same annotated MoA (compounds) or are part of the same gene family or targeted by the same compound (genetic perturbagens) and that have been shown to connect strongly to each other in CMap. Strong connectivity to a PCL offers a reductive but more interpretable view of connections because it represents connectivity to a group of related perturbagens rather than just a single perturbagen.
Consistent with our observation that a number of CDK inhibitors connect strongly to apigenin, we see that the top PCL is a set of CDK inhibitors known to connect to each other in CMap.
Now apply our hypothesis that apigenin functions as a CDK inhibitor to the genetic perturbagen data. For example, if apigenin is a cdk inhibitor, does it show a strong connection to knockdown of a CDK gene of genes? Use the Quick Tools menu to select the Gene Knock-Down perturbagens, then search for CDK in the rows and use the upward-facing arrow key to bring the selected rows to the top of the heatmap for easy viewing (those rows will now have a blue background). The image below shows a view of apigenin connections filtered for CDK genes. Knockdown of CDKL4 shows significant connectivity to apigenin (summary score = 91). CDK1 knockdown also shows connectivity across 3 cell lines.
A closer look shows that apigenin-treatment in MCF7 cells connects to knockdown of four CDK genes, particularly CDK9 and CDK7, suggesting a cell-line specific effect of this compound treatment.
Overall, these results suggest that apigenin has CDK inhibition activity and further experiments pertaining to this hypothesis are warranted.