Sciope

Sciope facilitates machine-learning-assisted model exploration and likelihood-free inference for discrete stochastic gene regulatory network models using scalable computational methods.

Key Features:

• Machine Learning-Assisted Methods: Incorporates advanced machine-learning algorithms to support model exploration and manage large-scale parameter sweeps.
• Likelihood-Free Inference: Supports likelihood-free inference methods for discrete stochastic gene regulatory networks, avoiding explicit likelihood calculations.
• Scalability and Parallelism: Implements a scalable architecture that utilizes distributed and heterogeneous computational resources to enable parallel parameter exploration across platforms from workstations to cloud.
• Discrete Stochastic Models: Targets discrete stochastic models of gene regulatory networks for simulation and analysis.
• Parameter Exploration and Optimization: Enables large-scale parameter sweeps and systematic parameter exploration for model discovery and optimization.
• Extensibility for Algorithms: Provides an architecture to implement and test novel machine-learning-assisted algorithms for model exploration.
• Implementation: Packaged as a Python 3 library.

Scientific Applications:

• Gene Regulatory Networks: In silico studies and systematic investigation of stochastic gene regulatory networks.
• Model Exploration and Validation: Extensive parameter exploration to develop and validate theoretical models against experimental data.

Methodology:

Implements machine-learning-assisted model exploration and likelihood-free inference, supports large-scale parameter sweeps, and provides a scalable distributed architecture enabling parallelism and extensibility for new algorithms.