GRAND PRIZE
July 15, 2003 | Identifying cellular pathways that are active or perturbed in a biological experiment is a key task within drug discovery, such as the validation of a target or the study of a compound mechanism of action. In the past decade, microarray transcript profiling (TP) has been developed into a powerful method for the identification of perturbed cellular processes.
But TP has its shortcomings. First, it can make statements only about genes that are measurably regulated at the transcriptional level. Second, the measurement of these variations suffers from noise inherent to the technology. Third, even when clustering algorithms can be used to group genes and reduce readout complexity, it is still tedious to derive associations to the perturbed pathways.
Proteins work sequentially along pathways and also in a coordinated manner inside functional modules such as the RNA polymerase complex. Such knowledge, when available in a computable format, can be used to overcome the shortcomings mentioned above:
· It can help detect the potential activity of a protein that's not transcriptionally regulated. The activity of a transcription factor can be suspected because of the differential expression of its targets.
· The variance of the experimental noise is reduced when working with average values over groups of genes. The weak regulation of a single subunit in a complex could be noise. The weak, but simultaneous, regulation of all subunits can be statistically significant.
| As the diversity and complexity of the data available to life scientists increases, so does the difficulty of integrating and interpreting those data. When thousands of genes, for example, are differentially expressed in a microarray experiment, translating the observed changes in terms of perturbed cellular processes can be a tough problem. To address this and other high-throughput-data challenges, Millennium Pharmaceuticals built the PAthway Resource and Information System, or PARIS — a unique platform for combining knowledge from heterogeneous data sources in the construction of a pathway knowledgebase. With core technology from Ingenuity Systems, Millennium has developed a platform capable of integrating gene, protein, cell, small molecule, and disease data into a unified framework of understanding centered upon a common pathway context.
"This kind of fusion of life science data is one of the most challenging problems facing the pharmaceutical community, as data is just data unless there's a context around which knowledge can be extracted," notes John Reynders, vice president of information systems at Celera Genomics, and a judge for the Discovery and Development category of Bio·IT World's Best Practices Awards.
— Editors
|
For these reasons, Millennium Pharmaceuticals developed the PAthway Resource and Information System (PARIS), consisting of the Activity Center scoring algorithm for analyzing transcriptional data, and a visualization tool for display of the analysis results. The analysis algorithm is based on a graph of metabolic and signaling protein networks, which was derived from the public databases LIGAND and CSND, as well as Ingenuity's Pathways Knowledge Base.
Ingenuity's Pathways KB provides the Millennium team with structured pathway knowledge to power the PARIS analysis. The Pathways KB is the world's largest curated, computable model of biological pathways — a network of millions of individually modeled interrelationships among proteins, cells, small molecules, and diseases.
The approach of using information from Ingenuity's knowledgebase and applying scoring algorithms on biological data allows several important applications, including personalized medicine and biological marker discovery. For example, PARIS revealed global differences in multiple myeloma biology between groups of patients who showed different responses to Millennium's proteasome inhibitor Bortezomib (Velcade). Transcriptional profiles of myeloma cells were analyzed for pharmacogenomic efficacy marker discovery. In addition, the comparison between the two patient populations revealed that integrin and adhesion, tumor necrosis factor (TNF) signaling, cell cycles, and interleukin signaling pathways accounted for some of the overall difference.
Adapted from company entry