Summary

The fields of pharmacology, biotechnology and chemistry are becoming increasingly reliant on high throughput screening (HTS). HTS is a method to perform systematic large numbers of automated biochemical
experiments. A newer development of HTS is high content screening (HCS). In HCS experiments, images are taken from live or fixed cells to see phenotypic changes after admission of a biochemical reagent. The scale in which images are created using HCS makes it implausible to manually review all these images. Using a method named cellpainting, numeric datasets are generated describing the information of the images.
Big data analyses platforms like StratoMineR, Tibco Spotfire and Dotmatics exist to analyse these datasets. Whilst numeric data analysis is proven to create credible results, we hypothesise that higher quality results can be achieved by reuniting the images with numeric data. To test this hypothesis we have asked the following questions: (MRQ) “What is the value of uniting HCS image data with numeric data?” (sq1) “Does this connection aid in curating labels, eliminating extreme outliers thus increasing the quality of training data?” (sq2) “Can this connection add value to the verification and confirmation process of promising hits?”