A computer algorithm has been developed to effectively identify genetic mutations that interact with each other to promote cancer, alongside other critical genetic insights that researchers hope to leverage for creating new cancer treatments in the future. This research team utilized an innovative network computer model to discover mutations that occur together and other similarities in DNA sequences across various cancer types. This model simplifies the quest to find patterns within vast datasets of cancer genetic information.
A computer algorithm has been developed to effectively identify genetic mutations that interact with each other to promote cancer, alongside other critical genetic clues that researchers hope to leverage for creating new cancer treatments in the future.
As detailed in the journal Frontiers in Bioinformatics, a team from Washington State University created a unique network computer model that finds co-occurring mutations and identifies similarities in DNA sequences across multiple cancers. This model facilitates easier pattern searches in extensive cancer genetic datasets.
“This is not a study focused on just one type of cancer; it’s an analysis of many cancers at once, searching for patterns that could eventually aid in drug discovery,” said Assefaw Gebremedhin, the study’s corresponding author and an associate professor in WSU’s School of Electrical Engineering and Computer Science.
Cancer is often perceived as a single disease, but in reality, it comprises a wide range of diseases, each driven by different mutations that influence disease progression and patient outcomes, explained co-author Steven Roberts, a molecular genetics researcher at the University of Vermont. Gaining insights into how prevalent various mutations and driver genes are across different cancers can help prioritize treatment targets. However, researchers have faced challenges due to the immense computational demands of analyzing extensive genetic sequences and numerous mutations.
“We couldn’t analyze all the sequences because it would overwhelm the computational capacity,” remarked Roberts. “Attempting to combine and evaluate all genomic data creates a mathematical problem that grows exponentially, which can easily overload the system.”
The WSU team’s network model, named DiWANN, is more efficient and less cluttered than existing models, yet it preserves crucial structural components.
“This model represents the simplest way to depict information without losing any critical details,” noted Gebremedhin. “Our approach focuses on understanding relationships between sequences more effectively, which means faster computations. By using a more minimalistic representation of the network, we can extract more insights while making computations manageable.”
Since the inception of their DiWANN model five years ago, the researchers have applied it to investigate the geographical patterns of tick-borne illnesses and the transmission of COVID-19 during the pandemic.
In their recent study, the researchers introduced an additional data reduction step to lessen computational requirements and employed a secondary computer model to enhance their understanding of co-occurring genes. This effort, led by WSU computer science Ph.D. student Shruti Patil, marks her as the study’s first author.
The findings indicated a consistent occurrence of two mutations in pancreatic cancer, which was previously only hypothesized. One mutation, known as tumor protein 53, serves to inhibit tumor growth, while the other, KRAS, promotes cellular proliferation.
The team identified cancer types that have strong connections to one another and may respond to shared drug treatments. Some cancers were found to be homogeneous, sharing similar mutations that lead to the disease, while others exhibited a vast diversity of mutations.
“Certain cancer types show a high level of similarity, with their driver mutations frequently being identical, whereas in other types, the mutations are quite diverse,” Roberts said. “Based on our predictions, those varied cancers may prove to be more challenging to treat.”
Thanks to the sparse information provided by the WSU network model, researchers can broaden the scope of tumors for their studies.
“This enhances our ability to uncover new interactions and insights into tumor behavior,” Roberts stated. “Being able to rapidly screen through large datasets presents a much more efficient approach.”
The team is currently developing an online tool for public health experts, enabling researchers in the health sector to utilize the model for exploring complex issues related to cancer and other diseases. This research received backing from the National Science Foundation and the National Cancer Institute.
