Researchers at the prestigious Indian Institute of Technology Madras (IIT-M) have developed a machine learning-based computational tool to better detect cancer-causing tumors of the brain and spinal cord. A web server known as “GBMDriver” (GlioBlastoma Mutiforme Drivers) is available online.
GBMDriver was specifically developed to identify driver and passenger mutations (passenger mutations are neutral mutations) in glioblastoma. Various factors such as amino acid characteristics, di- and tri-peptide motifs, conservation scores, and position-specific scoring matrix (PSSM) were considered to develop this web server.
In this study, 9,386 driver mutations and 8,728 passenger mutations in glioblastoma were analyzed. Glioblastoma driver mutations were identified with 81.99% accuracy in a blind set of 1809 variants. This is superior to existing calculation methods. This method is completely dependent on the protein sequence.
Read also | Research team at IIT Guwahati develops catalyst to produce hydrogen from wood alcohol
Glioblastoma is a rapidly growing tumor of the brain and spinal cord. Although research is being conducted to understand this tumor, treatment options remain limited and survival rates of less than two years from initial diagnosis are expected, IIT-M said.
The research was led by Professor M. Michael Gromiha of the Department of Biotechnology at IIT-M, and the findings were published in the respected peer-reviewed journal Briefings in Bioinformatics.
“We identified key amino acid signatures for identifying cancer-causing mutations and achieved the highest accuracy for distinguishing between driver and neutral mutations. We hope it will help prioritize glioblastoma driver mutations, help identify potential therapeutic targets, and help develop drug design strategies,” said Professor Gromiha.
Major applications of this study include transplantable methodology and features to apply to other diseases, which may serve as one of the important criteria for disease prognosis.
“Our method showed precision and AUC of 73.59% and 0.82, respectively, on 10-fold cross-validation, and 81.99% and 0.87 on a blind set of 1809 variants. We believe that this will help us prioritize driver mutations and identify therapeutic targets,” said Medha Pandey, a PhD student at IIT-M.
