MVIAm

MVIAm applies multi-view integrative analysis to microarray gene expression data to identify significant biomarkers and classify cancer samples using cross-platform normalization and Multi-View Self-Paced Learning (MVSPL).

Key Features:

• Cross-platform normalization: Applies various cross-platform normalization methods to combine multiple microarray datasets into a cohesive multi-view dataset.
• Multi-View Self-Paced Learning (MVSPL): Uses MVSPL as a gene selection mechanism tailored for cancer classification problems.
• Machine learning integration and classification: Leverages advanced machine learning techniques for integration, classification, and identification of significant biomarkers across datasets.
• Robustness to data challenges: Targets high noise, high dimensionality with small sample sizes, batch effects, and low reproducibility in microarray gene expression analysis.
• Validation on simulated and real data: Demonstrated efficacy using simulated data and real-world breast and lung cancer microarray datasets.

Scientific Applications:

• Biomarker identification: Identification of significant biomarkers from integrated microarray gene expression datasets.
• Cancer classification: Gene selection and classification in cancer studies, including breast and lung cancer.
• Cross-platform integrative analysis: Integration of cross-platform microarray datasets to mitigate batch effects and improve reproducibility of results.

Methodology:

Combine datasets via cross-platform normalization to form a multi-view dataset, apply Multi-View Self-Paced Learning (MVSPL) for gene selection and classification, and evaluate on simulated and breast/lung cancer microarray datasets.