Gene-Specific Machine Learning Models to Classify Driver Mutations in Clonal Hematopoiesis

Area of Science:

• Genetics
• Computational Biology
• Hematology

Background:

• Clonal hematopoiesis, age-related stem cell expansions, is driven by mutations lacking consensus classification.
• Current methods for variant classification rely on expert-derived rules, limiting objective analysis.

Purpose of the Study:

• To develop and validate machine learning models for classifying somatic mutations in blood.
• To identify driver mutations in clonal hematopoiesis using a purely data-driven approach, independent of prior knowledge.

Main Methods:

• Trained machine learning models on mutation data from 12 key genes.
• Validated model performance in classifying somatic mutations relevant to clonal hematopoiesis.
• Employed a data-driven methodology, avoiding pre-existing assumptions about driver mutations.

Main Results:

• Successfully trained and validated machine learning models capable of classifying somatic mutations.
• Demonstrated a purely data-driven method for identifying potential driver mutations in clonal hematopoiesis.
• The models provide a novel, objective classification of mutations based on observed data.

Conclusions:

• Machine learning offers a powerful, data-driven alternative to expert-based classification of clonal hematopoiesis driver mutations.
• This approach can enhance the understanding and identification of mutations driving age-related stem cell expansions.
• Future research can leverage these models for more accurate and objective genetic variant analysis in hematology.