Introduction

This repository contains the official implementation of the paper: "Genome-wide mining and de novo designing of plant enhancers".

Installation

First, clone the repository:

git clone https://github.com/shiyegao/BaseSearch.git
cd BaseSearch

Then, install the uv package manager.

Once uv is installed, install the project dependencies:

uv sync

Usage

Regressor Training

To train enhancer activity regressors, run the following commands:

uv run adaboost_reg.py
uv run rf_reg.py
uv run convnet_reg.py

You can modify the model hyperparameters in the corresponding configuration files located at: conf/regression.

Enhancer Design

To accelerate the design process, you can reduce the num_iterations parameter in the config file: conf/design.

[Optional] Wandb

If you’d like to log experiments with wandb, set the use_wandb to True in the config file. Also, specify your wandb_proj_name and entity.

BaseSearch Algorithm

Run the BaseSearch algorithm with:

uv run design.py conf/design/basesearch.py

Baseline Methods

You can also run several baseline enhancer design strategies:

uv run design.py conf/design/random.py
uv run design.py conf/design/random_single.py
uv run design.py conf/design/beam.py
uv run design.py conf/design/anneal.py
uv run design.py conf/design/mcts.py
uv run design.py conf/design/ga.py