Identification of Clonal Hematopoiesis Driver Mutations through In Silico Saturation Mutagenesis

Area of Science:

• Genetics and Genomics
• Computational Biology
• Hematology

Background:

• Clonal hematopoiesis (CH) involves the expansion of hematopoietic stem cells due to somatic mutations, increasingly linked to hematologic malignancies and cardiovascular diseases.
• While common CH driver genes are known, a comprehensive understanding of mutations initiating CH is lacking.
• Existing methods for identifying driver mutations are limited.

Approach:

• Developed high-quality machine learning models for 12 recurrent CH driver genes to identify initiating mutations.
• Validated model performance against experimental base-editing and expert-curated rules.
• Applied models to a large dataset (UK Biobank) to identify CH driver mutations.

Key Points:

• Machine learning models demonstrate superior performance in identifying CH driver mutations compared to experimental and rule-based approaches.
• The models successfully identified CH driver mutations in nearly half a million UK Biobank donors.
• Findings confirm known associations between CH driver mutations, age, and disease prevalence.

Conclusions:

• The developed machine learning models offer a robust tool for accurate CH driver mutation identification.
• These models can facilitate the accurate identification of clonal hematopoiesis in healthy populations.
• This work advances the systematic landscape of mutations capable of initiating clonal hematopoiesis.