Depick Knows Where Drugs Dock

Cell-based test systems have proven satisfactory for identifying new pharmacologically active molecules. Scientist from the Helmholtz Zentrum München use DePick software to show how mathematics can be used to find the right target structures.

DePick allows to start with phenotypic characteristics to identify target structures that are affected by small molecules. Source: Dr. Mónica Campillos / HMGU

Researchers use a cell-based test system to examine libraries of thousands of molecules in order to develop new medicinal products. They use color reactions to select molecules with the required characteristic.

These so-called phenotypic cellular approaches have a crucial advantage over target-based models.* Primarily, scientists can better examine the effect of molecules on disease processes. The challenge that remains is to use the phenotype as a basis for reaching conclusions regarding the target structures affected by the pharmaceuticals.

Tracking down correlations with DePick

“We developed DePick in order to solve this problem,” Dr. Mónica Campillos says. She directs the “Systems Biology of Small Molecules” junior research group in the Institute of Bioinformatics and Systems Biology (IBIS) at the Helmholtz Zentrum München. Further project partners were the Comprehensive Pneumology Center at the Helmholtz Zentrum München and the National University of Singapore.

DePick carries out a so-called deconvolution. “This is an approach that is also used in signal processing and image processing,” Campillos explains. The objective is to use a mathematical transformation to determine factors underlying observations. Campillos continues: “DePick allows us to start with phenotypic characteristics to identify target structures that are affected by small molecules.”

Together with colleagues under the leadership of Dr. Andreas Ruepp (IBIS) and Melanie Königshoff, MD, PhD, head of the research unit “Lung Repair and Regeneration” at Helmholtz Zentrum München, the research group of Mónica Campillos examined eight publicly available test systems.” They succeeded in making 59 associations between phenotype, drug and target, including one previously unidentified connection between the Wnt signaling pathway** and the aromatase CYP19A1***. “The example shows how a combination of mathematical methods with tests that are based on a phenotype makes it possible to demonstrate previously unknown links,” Königshoff remarks.

Further information

Publication: Liu, X. et al., Systematic identification of new pharmacological targets from small molecule phenotypic screens. Cell Chemical Biology, DOI:

DePick is available as a download online service: (

*Target refers to a biological structure that is affected by drugs. Suitable molecules are developed on the computer as soon as the target’s structure is known.

**The Wnt signaling pathway is one of many pathways for the transmission of signals that allow cells to respond to external changes. The signaling pathway is named after its main player “Wnt”, a signaling protein that, as a local mediator, takes on a key function in the development of various animal cells. Numerous proteins are involved in transmitting the signals, including beta-catenin.

***The aromatase CYP19A1 is an enzyme that catalyzes the conversion of testosterone into estradiol and of androstenedione into estrone (aromatization). This is the last reaction step in the biosynthesis of the estrogens.