Making the connection for cell-signaling networks
hough McFarland-Bascom Professor Robert Nowak initially developed a model that could trace telephone networks for intelligence purposes, he learned recently that he could apply it in an area outside of information technology. In collaboration with Genetics Professor Audrey Gasch, Nowak is applying his model to a network that has long baffled biologists: cell protein signaling.
In living cells, networks or signaling proteins communicate information from the cell surface to the nucleus, prompting certain genes to turn “on” or “off.” Thus, cells can adapt to environmental changes. If scientists could understand and regulate this gene modulation, it could open new research opportunities into areas of human disease and cell growth, biosensor development and biofuel manufacturing.
Unfortunately, it is impossible to directly measure protein signaling interactions on a large scale in vivo. Current knowledge of cell signaling networks is limited to datasets gathered by cell-study technology and theory based on knowledge of protein reactions. “With our technique, we can mine that data and look at it collectively from a network-centric perspective,” says Nowak.
Currently, scientists can determine which proteins are involved in a pathway, but not their order. By applying his network tomography model, Nowak can reconstruct cell signaling networks from genomic datasets and show the most likely sequence of proteins in a pathway. The model also can show how closely certain proteins interact. For example, if two proteins appear together in a number of pathways, they are more likely to operate closely in sequence. “Roughly speaking, cell signaling networks are not very different from the telephone network or the Internet,” says Nowak.
Next, Nowak hopes to optimize the model by augmenting information from genomic datasets with knowledge from other cell-study methods.
 |
 |
Predicting protein signaling interactions in yeast. By using genomic network tomography (GNT) analysis, the number of potential inter-actions is reduced from 143 to 91.
(Larger image 1) (Larger image 2) |
