omniBiomarker

omniBiomarker integrates knowledge from the NCI Cancer Gene Index to guide selection of biologically relevant feature-selection algorithms for identifying cancer biomarkers from high-throughput genomic expression data and for improving microarray-based clinical predictions while mitigating instability from the curse of dimensionality.

Key Features:

• Curse of dimensionality mitigation: Addresses small-sample, high-feature scenarios in genomic expression data to reduce instability in biomarker selection.
• NCI Cancer Gene Index integration: Uses the NCI Cancer Gene Index as an external knowledge source to inform feature-selection choices.
• Biological relevance scoring: Computes the biological relevance of feature-selection algorithms based on an externally validated knowledge base of manually curated cancer biomarkers.
• Feature-selection algorithm guidance: Guides the selection of algorithms for feature selection to improve reproducibility of biomarker identification across algorithm choices.
• Microarray clinical prediction support: Focuses on improving the reliability and accuracy of microarray-based clinical predictions derived from genomic expression data.

Scientific Applications:

• Cancer biomarker discovery: Identification and prioritization of candidate cancer biomarkers from high-throughput genomic expression datasets.
• Feature selection and validation: Selection and biological validation of features from microarray and other genomic expression data.
• Reproducibility assessment: Evaluation and reduction of algorithm-induced variability in biomarker identification to improve reproducibility.
• Microarray-based clinical prediction enhancement: Support for developing more biologically grounded clinical prediction models using microarray data.

Methodology:

Integrates the NCI Cancer Gene Index and an externally validated, manually curated cancer biomarker knowledge base to compute biological relevance scores for feature-selection algorithms and to guide algorithm selection to mitigate instability from the curse of dimensionality.