OmicsMarkeR

OmicsMarkeR performs multivariate classification, feature selection and performance–stability evaluation on metabolomics, proteomics, and transcriptomics datasets to identify and quantify biomarkers for diagnosis, prognosis, and risk prediction.

Key Features:

• Multivariate classification techniques: Implements Partial Least Squares Discriminant Analysis (PLS-DA), Support Vector Machines (SVM), Random Forest, Regularized Generalized Linear Models, and Prediction Analysis for Microarrays.
• Feature selection stability: Incorporates metrics to evaluate the stability of selected features across resampling and algorithmic variations.
• Robustness-Performance-Trade-off (RPT): Evaluates and balances classification accuracy against feature selection stability using the RPT framework.
• Dataset benchmarking: Compares algorithm performance on both simulated and real datasets, including simulated metabolomics scenarios and open-access experimental datasets.

Scientific Applications:

• Biomarker identification and quantification: Facilitates selection and quantification of candidate biomarkers from omics-level data.
• Disease diagnosis, prognosis and risk prediction: Supports studies aimed at diagnosis, prognosis and risk prediction for conditions such as Alzheimer's disease, cancer, diabetes, and trauma.
• Metabolomics analysis: Targets metabolomics datasets specifically while remaining applicable to proteomics and transcriptomics comparisons.

Methodology:

Performs in-depth comparisons of algorithm performance on generated simulated datasets (independent null with no correlation or discriminatory variables; correlated null with no discriminatory variables; correlated discriminatory) and on three open-access real datasets (two Nuclear Magnetic Resonance (NMR) and one Mass Spectrometry (MS)).