MSINGB: A Novel Computational Method Based on NGBoost for Identifying Microsatellite Instability Status from Tumor...

Area of Science:

• Oncology
• Bioinformatics
• Computational Biology

Background:

• Microsatellite instability (MSI) is a key mutator phenotype in various cancers, driven by DNA mismatch repair deficiency.
• MSI serves as a critical biomarker for cancer diagnosis, prognosis, and guiding therapeutic selection.
• Current experimental methods for MSI identification are laborious, time-consuming, and expensive.

Purpose of the Study:

• To develop a more effective computational method for identifying MSI status.
• To compare the performance of various machine learning and deep learning models for MSI prediction.
• To identify key features related to MSI and understand their impact on model predictions.

Main Methods:

• Developed MSINGB, a method utilizing NGBoost (Natural Gradient Boosting) for MSI status identification.
• Evaluated 11 machine learning and 9 deep learning models for MSI prediction performance.
• Employed feature selection strategies and SHAP for model interpretability.

Main Results:

• NGBoost demonstrated superior performance among the 20 evaluated models for MSI identification.
• MSINGB achieved enhanced prediction accuracy on both cross-validation and independent test datasets.
• Identified a compact, relevant feature subset strongly associated with MSI status.

Conclusions:

• MSINGB offers a highly accurate and efficient computational approach for MSI status identification.
• The study highlights the potential of NGBoost and feature selection for improving MSI prediction.
• This method can aid in more precise cancer diagnosis, prognosis, and treatment selection.