Identification of Clonal Hematopoiesis Driver Mutations through In Silico Saturation Mutagenesis

Area of Science:

• Genetics
• Computational Biology
• Hematology

Background:

• Clonal hematopoiesis (CH) involves stem cell expansion due to somatic mutations.
• CH is associated with hematologic malignancies and cardiovascular diseases.
• A comprehensive understanding of CH-initiating mutations is currently lacking.

Purpose of the Study:

• To develop and validate machine learning models for identifying driver mutations in 12 recurrent CH genes.
• To compare the performance of these models against expert-curated rules.
• To apply the models to a large dataset for identifying CH mutations and their associations.

Main Methods:

• Training gene-specific machine learning models for 12 key CH genes.
• Evaluating model performance against established expert-curated rules.
• Applying validated models to UK Biobank data (nearly 500,000 donors).

Main Results:

• Machine learning models demonstrated superior performance in identifying CH driver mutations compared to expert rules.
• Application to UK Biobank data successfully identified CH driver mutations.
• The study reproduced known associations between CH mutations, age, and disease prevalence.

Conclusions:

• Developed and validated machine learning models offer an accurate method for identifying CH driver mutations.
• These models provide an advantage over existing expert-curated rules.
• The models can significantly aid in the identification and clinical interpretation of CH mutations in individuals.