RED

Deep Reinforcement Learning for Optimal Experimental De# Biology

Installation

RED does not need to be installed to run the examples

To use the package within python scripts, RED must be in PYTHONPATH.

To add to PYTHONPATH on a bash system add the following to the ~/.bashrc file

export PYTHONPATH="${PYTHONPATH}:<path to RED root dir>"

Dependencies

Standard python dependencies are required: numpy, scipy, matplotlib. TensorFlow and hydra-core are required). Instructions for installing 'TensorFlow' can be found here: https://www.tensorflow.org/install/

User Instructions

Code files can be imported into scripts, ensure the RED directory is in PYTHONPATH and simply import the required RED classes. See examples.

To run examples found in RED_master/examples from the command line, e.g.:

$ python train_RT3D_prior.py 

The examples will automatically save some results in the directory:

The main classes are the continuous_agents and OED_env, see examples for how to use these:

continuous_agents

The continuous_agents.py file can be imported and used on any RL task.

from RED.agents.continuous_agents import RT3D_agent

OED_env

Contains the environments used for RL for OED. Can be imported and initialised with any system goverened by a set of DEs

from RED.environments.OED_env import OED_env