The predominant strategies for drug discovery today employ high-throughput screens and microarray-based methods that generate large data sets. Increasingly, these techniques are being used to study biological processes as a function of time and/or chemical concentration. Functionally-related processes may vary in timecourse or strength in different experiments or individuals. Current clustering methods do not highlight their similarities, because while their expression- or reaction profiles may share structural features, they may be compressed or attenuated relative to each other in duration or amplitude. Extracting maximal information from complex data arrays requires methods that can "fit" curves representing the different biological events (i.e., gene or protein expression/activity) by compensating for distortions caused by variations in rate or abundance. Harvard Medical School researchers have developed dynamic software that is capable of addressing this important need.

The software is superior to other available clustering techniques for anlyzing time and concentration data series, because it functions independent of differences in absolute rate or amount of change in the test parameter from one such series to another. It is flexible, in that it corrects phase shifts between expression/reaction profiles and identifies functionally-related processes through time-independent comparison of expression/reaction profiles. The program also is versatile. While it is optimized for analysis of protein or RNA expression data, it also is adaptable to large arrays of complex parameters, i.e. phenotypic data in clinical trials.