iCancer-Pred

iCancer-Pred predicts cancer types from DNA methylation patterns to classify seven cancer types based on epigenetic markers.

Key Features:

• Data Source: Uses DNA methylation datasets for seven cancer types obtained from The Cancer Genome Atlas (TCGA).
• Feature Reduction and Selection: Applies coefficient of variation for initial feature reduction followed by an elastic network for feature selection.
• Predictive Model: Trains a fully connected neural network on the selected methylation features for cancer identification and classification.

Scientific Applications:

• Cancer classification: Predicts seven cancer types with overall accuracy exceeding 97% as evaluated by 5-fold cross-validation using DNA methylation features.

Methodology:

Collect DNA methylation data from TCGA for seven cancer types; apply coefficient of variation for feature reduction; use an elastic network for feature selection; build and train a fully connected neural network on the selected features.