Identification of potential driver mutations in glioblastoma using machine learning

Abstract

Glioblastoma is a fast and aggressively growing tumor in the brain and spinal cord. Mutation of amino acid residues in targets proteins, which are involved in glioblastoma, alters the structure and function and may lead to disease. In this study, we collected a set of 9386 disease-causing (drivers) mutations based on the recurrence in patient samples and experimentally annotated as pathogenic and 8728 as neutral (passenger) mutations. We observed that Arg is highly preferred at the mutant sites of drivers, whereas Met and Ile showed preferences in passengers. Inspecting neighboring residues at the mutant sites revealed that the motifs YP, CP and GRH, are preferred in drivers, whereas SI, IQ and TVI are dominant in neutral. In addition, we have computed other sequence-based features such as conservation scores, Position Specific Scoring Matrices (PSSM) and physicochemical properties, and developed a machine learning-based method, GBMDriver (GlioBlastoma Multiforme Drivers), for distinguishing between driver and passenger mutations. Our method showed an accuracy and AUC of 73.59% and 0.82, respectively, on 10-fold cross-validation and 81.99% and 0.87 in a blind set of 1809 mutants. The tool is available at https://web.iitm.ac.in/bioinfo2/GBMDriver/index.html. We envisage that the present method is helpful to prioritize driver mutations in glioblastoma and assist in identifying therapeutic targets.

States of Arts and New Approaches to Spinal Cord Tumors

“…Beyond diagnostic neuroradiology, molecular genomics is increasingly employed to decode the complexities of spinal cord tumors. Research, such as the work of Jung et al analyzing clinical and radiological data to predict H3 K27M mutations, and Pandey et al developing techniques to differentiate between driver and passenger mutations in glioblastoma, demonstrates the potential of these approaches [39,40]. These methods help prioritize essential mutations and inform the development of targeted treatments.…”

“…Radiotherapy is typically reserved for situations where en bloc resection is unfeasible. Traditionally, high doses of radiation (40-60 Gy) were required, leading to a high incidence of complications due to the proximity of the spinal cord and thoraco-abdominal organs, including radiation myelopathy and various issues affecting gastrointestinal and reproductive healthhormonal imbalances, reduced fertility, uterine dysfunction, miscarriage, preterm labor, low birth weight, and placental abnormalities [39,40,78]. However, with the advent of intensity-modulated radiation therapy and stereotactic radiosurgery, it's now possible to deliver high radiation doses directly to the spinal region while sharply reducing exposure to surrounding areas, thereby minimizing the side effects typical of conventional radiation treatments [79,80].…”

Current Applications of Machine Learning for Spinal Cord Tumors

“…Their random forest classifier showed that this histone gene mutation could be predicted with moderate discrimination (63.4% accuracy) based on clinical and radiological features. Pandey et al utilized specific peptide motifs, conservation scoring schemes, Position Specific Scoring Matrices, and mutation matrices to develop a machine learning method (GlioBlastoma Multiforme Drivers) distinguishing between driver and passenger mutations in glioblastoma based on recurrence in patients with an accuracy of 73.6% [ 22 ]. Such methods can be applied to prioritize driver mutations and facilitate the identification of therapeutic targets.…”

“…Overall, further studies are required to investigate prognostic biomarkers and targetable genetic drivers in order to translate them into clinical practice ( Table 1 ). [ 14 , 15 , 16 , 18 , 19 , 20 , 21 , 22 , 25 , 30 , 31 , 48 ].…”

Glioblastoma modeling with 3D organoids: progress and challenges

“…Several mutations or mutation combinations appeared to induce a large tumorigenic clonal expansion of the transduced cells in cerebral organoids, such as Myc overexpression, NF1/P53/PTEN triple knockout and CDKN2A/PTEN double knockout together with EGFRvIII overexpression. Interestingly, they found that Myc-overexpressing cells had a strong primitive neuroectodermal tumor-like signature with distinct cell identities compared with organoids carrying other mutation combinations (GBM-like neoCORs) [ 20 ] commonly found in GBM [ 65 , 72 , 92 ]. The GBM-like neoCORs showed not only an invasive and proliferative nature after renal subscapular engrafting but also active transcription of invasion-related genes including EMT-related transcription factors, proteases and migration-related receptors [ 20 ].…”