Visualization aims to provide a specific view of your data, with the hope that such a view leads to some form of understanding or insight. However, a given data set can be visualized in multiple ways, with each way possibly highlighting different aspects. When a tool (or a method) provides a single view of the data, there is the possibility that you might end up with a skewed view of what the data represents and thus make erroneous or incomplete conclusions. This talk will highlight how it is important to examine data in multiple ways, using datasets from high throughput screening and chemical library design as examples. First, I will highlight some well known examples of visualization methods that can hide important details of the underlying data such as barcharts, and methods such as binning that can also hide important distributional details. Second I will describe how alternate representations can provide a different view of the data. In particular I will describe a network based representation of sets of molecules, that can be used to compare libraries. The use of such a representation allows us to consider library contents and compare contents of two libraries using network metrics, thus providing an alternative to traditional similarity based methods. I will conclude with a summary of best practices that helps one maximize their understanding of their data.

How a visual vocabulary defines what you see in your data

2019 Fall National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.