Improved Discrimination of Disease States Using Proteomics Data with the Updated Aristotle Classifier

Area of Science:

• Biomedical data science
• Proteomics and metabolomics analysis
• Computational biology

Background:

• Mass spectrometry omics data are valuable for disease discrimination and biomarker discovery.
• Challenges in omics data analysis include missing values and a high feature-to-sample ratio, hindering machine learning applications.
• Accurate classification of disease states from omics data is crucial for clinical applications.

Purpose of the Study:

• To modify and optimize the Aristotle Classifier for improved performance on omics data.
• To develop a new supervised classification tool, AC.2021, for enhanced disease state identification.
• To provide researchers with a tool for more accurate classification of omics datasets.

Main Methods:

• Modification of the Aristotle Classifier into a new version, AC.2021.
• Benchmarking AC.2021 against its predecessor and established tools like Support Vector Machine (SVM) and XGBoost.
• Validation using multiple omics (proteomics) datasets to assess classification accuracy.

Main Results:

• The AC.2021 classifier demonstrated superior performance compared to existing methods, including SVM and XGBoost, on multiple datasets.
• AC.2021 effectively addressed challenges associated with missing data and high dimensionality in omics datasets.
• The optimized classifier achieved higher accuracy in discriminating disease states using proteomics data.

Conclusions:

• AC.2021 represents a significant advancement in machine learning for omics data analysis.
• The developed classifier offers improved accuracy for identifying disease states and biomarkers.
• The provided code enables researchers to leverage omics data more effectively for clinical insights.