DREAMTools

DREAMTools evaluates scoring metrics for DREAM challenges to enable rigorous assessment of computational methods in systems biology and translational medicine.

Key Features:

• Python package: Implements scoring and evaluation functionality for DREAM challenge datasets.
• Scoring metric evaluation: Implements evaluation of scoring metrics used in DREAM challenges.
• Scoring framework: Provides a framework to score new challenges and ensure consistency in evaluations.
• Historical dataset coverage: Covers more than 80% of completed DREAM challenges as of March 2016, enabling benchmarking against historical datasets.
• Method benchmarking: Facilitates comparative evaluation of statistical, machine learning, and biophysical approaches.
• Data platform integration: Integrates with DREAM and Synapse challenge data and metadata sources.

Scientific Applications:

• Scoring metric validation: Validates and compares scoring metrics used in community-driven DREAM challenges.
• Benchmarking predictive models: Benchmarks predictive models for predicting unobserved outcomes and inferring unknown model parameters.
• Comparative method evaluation: Enables comparative assessment of statistical, machine learning, and biophysical methods to refine predictive approaches.
• Method validation: Validates novel computational methods against historical DREAM challenge datasets.

Methodology:

Computational evaluation of scoring metrics and scoring frameworks applied to DREAM challenge datasets, with comparative assessments across statistical, machine learning, and biophysical methods.