Can computers find a cure for cancer?

Understanding brain cancer

Brain cancer is one of the most poorly understood diseases in the medical profession, partly because we know so little about the brain and partly because there are relatively few cases of it: roughly 20,000 cases are reported per year. Even if it were possible to accumulate data on all 20,000 cases, the cancer is heterogeneous: it varies from patient to patient.

At the Swedish Medical Center in Seattle, and specifically at the Center for Advanced Brain Tumor Treatment (CABTT), raw processing power is required to analyse the genes in cancer patients and determine their genetic makeup. At that point, doctors can diagnose a treatment that is geared specifically to that patient. The more data collected, the better the treatment plan.

Using a 'next-generation genome sequencer', the researchers are able to map three million genomic expressions using desktop computers. Data models run as high as nine terabytes. In a sense, the process is applying programming techniques to biological programming: researchers analyse data that has malfunctioned at the molecular level and try to unlock the puzzle pieces that are causing a tumour or cancerous cells to spread in the brain.

Dr Anup Madan and Dr Greg Foltz are the co-founders of CABTT. Their research is one of comparison: seeing how a gene sequence in a normal brain compares to the gene sequence in a brain cancer patient. (They also use a map of a mouse brain that was generated at the Allen Institute for Brain Science.) Foltz says that brain cancer has a one- to two-year survival rate; yet only a handful of researchers are working on a cure for it.

Located in Seattle, the centre has easy access to several other cancer research institutions and is near to Microsoft and other high-tech firms. Microsoft has a Health and Life Sciences division that helps researchers understand .NET frameworks and assists with programming models.

Collaboration is close, and both the Allen Institute and Microsoft have set up social networking systems where researchers can compare findings – sometimes even annotating data at a molecular or genomic level. "We are being moved to this position in science where the traditional approaches do not work. There are targeted drugs in the pipeline but you could not do enough trials," says Foltz. "The traditional approach is to do an MRI, start the chemotherapy, perform surgery and use cancer drugs. The challenge is to use the computer to tune the treatment according to the genetics of the patient."

What's most interesting about the gene sequencing tests is that to the human eye, all tumour data looks the same. Yet the computer knows the exact differences between two tumours, which helps it to aid medical professionals when finding cancer therapies.

Although most of the medical research has moved to computerised analysis, the researchers are also aware of a basic axiom: the starting data used must be sufficient for the desired result. In the end, the answer is in the genes and molecules of the human body.

What's needed is a computerised system that can understand our bodies completely. As work on this science is progressing, we may yet see this goal achieved.


First published in PC Plus, issue 278