Abstract

The screening of new medical drugs needs high throughput test systems to analyze binding interactions, e.g. between test substances and molecular targets occurring in the human body. So called small-molecule microarrays are a means for automatic compound screening and discovery. In comparison to other microarray techniques, small-molecule microarrays suffer from varying fixing characteristics of the used substances.

Irregularities like blurred immobilization areas cause strong variations in dimension and shape of a microarray spot and a computer-based analysis is not possible reliably. Thus, each microarray needs to be carefully revised by a scientist. In this paper, we propose a new method based on image processing algorithms to detect spots on a microarray robustly, to quantify the spot intensity and to assign each spot a plausibility criterion with respect to crosstalk, background noise and image processing success.

The plausibility criterion is used as an acceptance measure for putative binders in a drug screening assay and decreases the dataset to a small subset of compounds to be revised.