NatDRAP

NatDRAP identifies Ras pathway activation from multiomic cancer data by integrating bulk RNA sequencing, copy number variation, and mutation data into a deep neural network optimized with a hybrid artificial bee colony and gradient-based optimization strategy for precision oncology analysis.

Key Features:

• Deep Neural Network (DNN) Model: Integrates bulk RNA sequencing, copy number variations, and mutation data from the PanCanAtlas across 33 cancer types for joint multiomic modeling.
• Nature-Inspired Optimization: Synergizes an artificial bee colony algorithm with gradient-based optimizers to collaboratively optimize DNN parameters.
• Robust Performance: Demonstrates superior performance versus benchmark methods in diagnosing Ras pathway aberrant activity across multiple cancer types.
• Downstream Analysis: Supports gene ontology enrichment and pathological analysis to characterize Ras pathway activation and related molecular mechanisms.

Scientific Applications:

• Precision oncology stratification: Identifies patients with Ras pathway alterations and hidden responders who may benefit from targeted therapies.
• Biomarker discovery: Enables discovery of novel biomarkers and therapeutic targets through integrated analysis of RNA-seq, copy number, and mutation data.
• Cross-cancer characterization: Facilitates comparative analysis of Ras pathway activation across 33 cancer types using PanCanAtlas data.

Methodology:

Integration of bulk RNA sequencing, copy number variation, and mutation data from PanCanAtlas into a deep neural network, with optimization via a hybrid artificial bee colony plus gradient-based optimizers, followed by gene ontology enrichment and pathological analysis.