My work on understanding the regulation of drug-metabolizing enzymes got me interested in the prediction of transcription factor binding sites. After 1998, the central role of nuclear receptors in the regulation of genes involved in drug metabolism became clear. These transcription factors can directly activated by binding to their ligands (which can be compounds that occur in the body, like hormones, compounds found in food, like cholesterol derivatives and fatty acids, or drugs), and then directly move to the cell nucleus, where they sit down on the DNA close to the genes they up-regulate. You could use the analogy that these are flags that indicate to the cell what gene products need to be made.

At the time, we were hunting for the special places in a cell’s DNA, where these transcription factors might bind. At the time, the feeling of leading experimentalists in the field was that these binding sites are best found experimentally (which is certainly true), or if at all by searching the DNA sequence by eye for characteristic patterns. This, I had a hard time to believe, and I started to think about an algorithm to detect the characteristic, bipartite motif of nuclear receptor binding sites.

Such started the development of NUBIScan, a computer program to detect those places in the DNA of our genome, where drug-activated transcription factors bind. With it, we performed studies that compared standard laboratory methods to detect transcription factor binding sites with a computer-driven approach, and found that application of the NUBIScan algorithm does not work all of the time, but tends to find false-positives places, but that the program still tremendously increases the efficiency of binding site analysis. The program is publicly accessible for non-commercial use: http://www.nubiscan.unibas.ch.

In this way, my path into bioinformatics started to unfold.

No Comments »