NCI Cancer Bulletin: A Trusted Source for Cancer Research NewsNCI Cancer Bulletin: A Trusted Source for Cancer Research News
March 1, 2005 • Volume 2 / Number 9 E-Mail This Document  |  Download PDF  |  Bulletin Archive/Search  |  Subscribe

NCI Cancer Bulletin Archive

Page Options

  • Print This Page
  • Print This Document
  • View Entire Document
  • Email This Document
  • View/Print PDF

The information and links on this page are no longer being updated and are provided for reference purposes only.

A Conversation with Dr. Javed Khan

Dr. Javed Khan is head of the Oncogenomics Section of NCI's Pediatric Oncology Branch. He trained at Cambridge University and came to the National Institutes of Health (NIH) in 1995. He joined NCI in 2001 and has made a number of contributions to the field of gene expression profiling while also focusing on the translation of these new discoveries into useful clinical tools. (See story.)

Dr. Javed Khan How does gene expression profiling work?
In brief, the idea is to see the forest for the trees. We have discovered that only about 3 percent of the human genome actually codes for some 25,000 genes that build specific proteins. How do you find which genes are on and which are off in cells? Even more importantly for clinicians, how does this affect the function of that cell?

We create small "chips" made of DNA fragments representing most of the genes in the human genome. In our new normal sample database, this means nearly 19,000 genes. Scientists then label the genes they are studying with fluorescence, and the DNA from those cells finds its match on the chip and lights it up. We know which gene it is because we built the coordinates of the chip map. Once you have a target organ or disease, you look for active genes with unusual expression profiles. We've built artificial neural networks that actually have "learned" to predict, correlating profiles to disease outcome better than the clinicians could.

Where do you see the field of oncogenomics headed?
I think the era of personalized medicine is on the horizon. Nanotechnology could one day replace the chemical nature of our microarrays with electronics, essentially hard-wiring a chip that would yield a definitive diagnosis from a small number of biopsy samples. If such nanodevices could be combined with sophisticated neural networks, they could be used both as diagnostic tools and as probes to look for specific biomarkers. Compared with current clinical approaches, these new nanodevices should be smart enough to yield profound results.