Poster Presentation 50th Lorne Proteins Conference 2025

Characterization of hTERT Missense Mutations via Machine Learning (#316)

Georgina Parra 1 , Yoochan Myung 1 , Qisheng Pan 1 , Tracy M Bryan 2 , Jessica K Holien 3 , David B Ascher 1
  1. University of Queensland, St Lucia, QLD, Australia
  2. Children’s Medical Research Institute, Faculty of Medicine and Health, University of Sydney, Westmead, New South Wales, Australia
  3. School of Science, STEM College, RMIT University, Bundoora, Victoria, Australia

Missense mutations in the human telomerase catalytic subunit hTERT, which is crucial for telomeric DNA repeats synthesis and elongation, have been associated with Telomere Biology Disorders (TBDs) and carcinogenesis. It is both time-consuming and technically challenging to experimentally elucidate the effects of all possible missense mutations. More importantly, current computational predictors are not TERT-specific and fail to capture its complex biological and structural context, as they primarily rely on sequence information. In this work, we conducted a qualitative computational analysis to identify potential pathogenic drivers of TERT and developed three machine learning models by integrating both sequence- and structure-based features accounting for the biological mechanisms of TERT. Compared to the state-of-the-art methods, our best-performing model achieved superior results with Matthew’s Correlation Coefficient (MCC) up to 0.613 on ClinVar and gnomAD curated variants. Further to that, our model was validated on a clinical dataset curated according to American College of Medical Genetics and Genomics/Association for Molecular Pathology (ACMG) guidelines with robust performance (AUC = 0.868). Feature interpretation revealed that TERT residue conservation, changes in hydrophobic and weak polar interactions, and the location of these mutations in the C-terminal Extension domain are critical determinants of pathogenicity, consistent with previous findings. Finally, the in-silico saturation mutagenesis was performed to present a mutational landscape of TERT, which could offer valuable insights into the molecular mechanisms driving TBDs and aid early diagnosis as well as personalized treatment strategies.