DEEPrior

DEEPrior predicts the oncogenic potential of gene fusions by analyzing the amino acid sequences of fused proteins to support cancer genomics research.

Key Features:

• Dual-Mode Functionality: Provides an Inference mode that predicts the likelihood that a specific gene fusion contributes to cancer and a Retraining mode that allows model refinement with new datasets.
• Sequence-Based Prediction: Uses the amino acid sequence of fused proteins as the primary input for prediction of driver versus passenger fusion status.
• Probabilistic Output: Produces a probabilistic assessment of oncogenic potential to enable prioritization of gene fusions across tumor types.
• Implementation: Implemented in Python 3.7 and based on deep learning methods.

Scientific Applications:

• Fusion Prioritization: Prioritizes candidate gene fusions for experimental validation in cancer genomics studies.
• Driver Identification: Assists in distinguishing oncogenic driver fusions from passenger mutations to inform downstream analyses.
• Target and Biomarker Discovery: Supports identification of potential therapeutic targets and biomarkers derived from predicted driver fusions.

Methodology:

Applies a deep learning model trained on amino acid sequences of fused proteins; Inference mode computes oncogenic likelihood and Retraining mode incorporates user-provided datasets; implemented in Python 3.7.