Computers Help Identify Hard-to-Find Cancer Genes
CREDIT: BioMedical | Shutterstock
There's still so much we don't know about cancer. But because this often-deadly disease manifests in our DNA, there’s an unfathomably large resource sitting right in our bodies: the human genome.
Using an algorithm specially designed to analyze the human genome for cancerous mutations, researchers from the Cancer Genome Atlas Research Network have assembled the most complete genetic profile yet of a type of blood cancer known as acute myeloid leukemia.
The algorithm, called Dendrix++, puts computers to work where the human brain falls short: It takes the genome, or genetic information, found in a patient's cancerous cells, and compares it to the genome of the patient's healthy cells. By analyzing the differences in the two data sets, the algorithm can identify the handful of genes among millions that are most likely triggering the cancer. Doctors can then focus on developing medicines that target and correct these specific genes.
In other words, researchers can and have used this algorithm to find the cancerous needle in the genetic haystack.
It’s no easy task. And further complicating the process, genetic mutation continues even after the original cancer-causing mutation occurs — so the algorithm had to be able to sort out which mutations were causing the cancer and which occurred as a result of the cancer spreading.
Gene mutation naturally occurs as people age, even in healthy people without any history of cancer.
In other words, the possible variations on this already-huge data set are so vast that an unassisted human observer couldn't possibly process it, much less identify patterns.
Using this algorithm, researchers at Washington University in St. Louis who are involved in the Cancer Genome Atlas Project have identified at least one sequence of genetic mutations, called "driver mutations," that could be the root cause of acute myeloid leukemia.
The researchers also found that the genomes of patients with this type of leukemia had an average of 13 mutations, which is lower than the average number of mutations found in adult cancers.
The full report, which has been made widely available in the New England Journal of Medicine, contains an enormous amount of data that will help medical researchers develop medicines targeted at correcting the problematic genes.
Dendrix++ is a variation of an algorithm called Dendrix that was specially modified to identify extremely rare mutations, such as the ones found in acute myeloid leukemia.
Dendrix and a similar algorithm, called HotNet, were developed by an interdisciplinary team of biologists and computer scientists at Brown University that included professors Eli Upfal, Ben Raphael and Fabio Vandin. The algorithms have been used in several prior cancer studies. In 2011, researchers used HotNet to identify several genetic sequences important in understanding and treating ovarian cancer.
“That's an advantage of computing: You can design an algorithm that can apply to more than just one set of data,” Upfal, a computer scientist, told TechNewsDaily.