CAMUR

CAMUR extracts multiple equivalent rule-based classification models from NGS RNA-seq gene expression data to broaden identification of genes associated with phenotypes such as cancer.

Key Features:

• Multiple Model Extraction: Iteratively computes numerous equivalent classification models instead of producing a single model, enabling identification of a broader spectrum of genes associated with the predicted class.
• Iterative Rule-Based Classification: Builds rule-based supervised classification models and repeats the procedure until a predefined stopping criterion is met.
• Power-set Computation and Elimination: Calculates the power set of genes present in extracted rules and systematically eliminates those gene combinations from the dataset to drive discovery of alternative models.
• Knowledge Repository and Querying: Stores extracted models and related data in a database and provides querying capabilities to retrieve and analyze stored models.

Scientific Applications:

• TCGA Breast Cancer analysis: Applied to TCGA RNA-seq data for breast cancer to generate multiple equivalent classification models and identify frequent genes and their associations.
• TCGA Head and Neck Cancer analysis: Applied to TCGA RNA-seq data for head and neck cancer to generate multiple equivalent classification models and identify frequent genes and their associations.
• TCGA Stomach Cancer analysis: Applied to TCGA RNA-seq data for stomach cancer to generate multiple equivalent classification models and identify frequent genes and their associations.

Methodology:

Iterative computation of rule-based classification models; calculation of the power set of genes involved in these rules; systematic elimination of the resulting gene combinations from the dataset; repeated classification until a predefined stopping criterion is satisfied.