The MIT-led researchers' new computer model may better identify mutations in cancer cells’ DNA that drive the growth of tumors. Photo by royaltystockphoto/Shutterstock
June 20 (UPI) -- A research team led by Massachusetts Institute of Technology has unveiled a new computer model to better identify mutations in cancer cells' DNA that drive the growth of tumors -- as a way to help target drug therapies.
Presently, at least 30% of cancer patients have no detectable driver mutation that can be used to guide treatment, MIT said in a news release about the modeling study published Monday in Nature Biotechnology.
While cancer cells can have thousands of mutations in their DNA, only a handful drive the progression of cancer, MIT said. The new computational tool, called Dig, is able to scan the entire genome of cancer cells rapidly and identify mutations that occur more frequently than expected, which would suggest they are driving tumor growth.
This type of prediction has been challenging because some regions of the genome have an extremely high frequency of neutral "passenger" mutations that drown out the signal of actual driver mutations, MIT said.
According to MIT, since the human genome was sequenced two decades ago, researchers have been searching for mutations that contribute to cancer by causing cells to grow uncontrollably or evade the immune system.
They have found certain mutations, including epidermal growth factor receptor, commonly seen in lung tumors, and BRAF, a common driver of melanoma, which can now be targeted by specific drugs.
While useful, such protein-coding genes comprise only about 2% of the genome, MIT said. The rest of the genome also contains mutations that can occur in cancer cells, but it has been much tougher to determine whether any of those mutations contribute to cancer development.
Currently, when patients' tumors are screened for cancer-causing mutations, a known driver will turn up about two-thirds of the time, MIT said. In the new study, the researchers found additional driver mutations in the genome that appear to contribute to tumor growth for an additional 5% to 10% of cancer patients.
Bonnie Berger, the study's senior author, cited "a lack of computational tools that allow us to search for these driver mutations outside of protein-coding regions."
Berger is the Simons Professor of Mathematics and head of the computation and biology group at the Computer Science and Artificial Intelligence Laboratory at MIT.
"That's what we were trying to do here: design a computational method to let us look at not only the 2 percent of the genome that codes for proteins, but 100 percent of it," Berger said in the MIT news release.
To accomplish this, the researchers trained a type of computational model known as a "deep neural network" to search cancer genomes for mutations that occur more frequently than expected.
First, they trained the model on genomic data from 37 different types of cancer, which allowed the model to determine the background mutation rates for each type.
Data used to train the model came from the Roadmap Epigenomics Project and an international data collection called the Pan-Cancer Analysis of Whole Genomes.
Researchers focused partly on so-called "cryptic splice mutations," which their modeling found seem to disrupt tumor suppressor genes, causing the cell to lose one of its defenses against cancer.
The number of cryptic splice sites that were found in this study accounts for about 5% of the driver mutations found in tumor suppressor genes.
MIT said targeting cryptic splice mutations may offer a new way to potentially treat certain patients, possibly by using an approach under development that uses short strands of RNA called antisense oligonucleotides to patch a mutated piece of DNA with the correct sequence.
Researchers said they also used their model to investigate whether common, already known mutations might drive different types of cancers.
They found, for example, that BRAF, linked to melanoma, also contributes to cancer progression in smaller percentages of other types of cancers, including pancreatic, liver and gastroesophageal.
"These results could help guide the clinical trials that we should be setting up to expand these drugs from just being approved in one cancer, to being approved in many cancers and being able to help more patients," said Maxwell Sherman, an MIT graduate student who is one of the lead authors of the study.
"We see our tool as a real game-changer," built upon a strong foundation of previous research revealing key properties of how tumors emerge and evolve, he said.
Moreover, he said, the cancer research community has been "diligent about applying whole genome sequencing to patient tumors and making this data widely available to study. Some efforts are now up to tens of thousands of samples, which is just an incredible resource."
He added: "We are excited to collaborate with these efforts to study the non-coding cancer genome with our tool. These large databases should provide the statistical power necessary for our tool to uncover numerous new insights about how cancers grow and how this impacts patients in the clinic."