Functional assessment of all ATM SNVs using prime editing and deep learning

Area of Science:

• Genetics and Genomics
• Cancer Biology
• Bioinformatics

Background:

• The Ataxia telangiectasia mutated (ATM) gene is crucial for DNA damage response.
• Loss of ATM function elevates cancer risk and impacts patient prognosis.
• Interpreting the functional significance of ATM variants, especially variants of uncertain significance (VUSs), is a significant challenge.

Purpose of the Study:

• To comprehensively assess the functional impact of all possible single-nucleotide variants (SNVs) in the ATM gene.
• To identify critical residues within ATM essential for its function.
• To develop predictive models for ATM variant pathogenicity to support clinical applications.

Main Methods:

• Experimental evaluation of 23,092 ATM SNVs using prime editing to assess cell fitness in the presence of olaparib.
• Analysis of cancer genetics data and UK Biobank data to correlate variant impact with clinical outcomes.
• Development and application of a deep learning model (DeepATM) to predict the functional effects of the remaining 4,421 ATM SNVs.

Main Results:

• Experimental validation identified critical residues within the ATM gene.
• The study successfully evaluated the functional impact of a large number of ATM SNVs.
• The DeepATM model demonstrated high accuracy in predicting the functional effects of previously unassessed ATM variants.
• Correlations were found between ATM variant impact and cancer risk/prognosis using real-world data.

Conclusions:

• This comprehensive functional evaluation of ATM variants provides valuable insights into DNA damage response and cancer predisposition.
• The developed DeepATM model offers a powerful tool for predicting the pathogenicity of ATM variants, supporting precision medicine.
• The study establishes a framework for addressing VUSs in ATM and potentially other genes, improving clinical decision-making.