The computational tool named ‘GBMDriver’ (GlioBlastoma Multiforme Drivers) has been developed by researchers from Indian Institute of Technology Madras (IIT Madras) to improve the detection of cancer-causing tumours in the brain and spinal cord using Machine Learning. This tool is now accessible to the public online. Glioblastoma, a rapidly growing tumour in the brain and spinal cord, is a challenging form of cancer with limited treatment options and an average life expectancy of less than two years after diagnosis, despite extensive research efforts.
Advancing therapeutic options for Glioblastoma patients requires the evaluation of the functional consequences of variants in proteins involved in the disease. However, identifying disease-causing mutations from the vast array of observed variants can be a daunting task. That’s where the GBMDriver tool comes in – it is specifically designed to differentiate between driver mutations and neutral passenger mutations in Glioblastoma. Access to the GBMDriver tool can be found at the link: https://web.iitm.ac.in/bioinfo2/GBMDriver/index.html.
The development of the web server involved the consideration of several factors such as amino acid properties, di- and tri-peptide motifs, conservation scores, and Position Specific Scoring Matrices (PSSM).
Through the analysis of 9386 driver mutations and 8728 passenger mutations in glioblastoma, the researchers were able to identify driver mutations with 81.99 percent accuracy in a blind set of 1809 mutants, surpassing existing computational methods. It is noteworthy that this method solely relies on the protein sequence.
Prof. M. Michael Gromiha from the Department of Biotechnology at IIT Madras led a research team that included PhD student Ms. Medha Pandey, and two IIT Madras alumni, Dr. P. Anoosha and Dr. Dhanusha Yesudhas, currently based in the US. Their research, which identified important amino acid features for detecting cancer-causing mutations, was published in the peer-reviewed journal, Briefings in Bioinformatics (https://doi.org/10.1093/bib/bbac451).
Ms. Pandey noted that their method achieved an accuracy of 73.59% and an AUC of 0.82 in 10-fold cross-validation, and an accuracy of 81.99% and an AUC of 0.87 in a blind set of 1809 mutants. She believes that this method can prioritize driver mutations in glioblastoma and aid in the identification of therapeutic targets.
