DAZZLE

This repository include code and documentation for our manuscript "Improving Gene Regulatory Network Inference using Dropout Augmentation".

Checkout RegDiffusion, our latest work on gene regulatory network inference: https://github.com/TuftsBCB/RegDiffusion

Install

This package is available on pip

pip install grn-dazzle

Basic Usage

The core function runDAZZLE requires the following two things to get started:

• Single cell gene expression table. We suggest you use log transformation to normalize the data
• Experiment Configs. We also provide two sets of default configs with this package, namely DEFAULT_DAZZLE_CONFIGS and DEFAULT_DEEPSEM_CONFIGS. They are just two python dictionaries. If you need to make modifications, just save them to a variable and adjust the values.

Quick Example

from dazzle import load_beeline, runDAZZLE, get_metrics, DEFAULT_DAZZLE_CONFIGS


bl_data, bl_ground_truth = load_beeline(
    data_dir='data', 
    benchmark_data="hESC", 
    benchmark_setting="500_STRING"
)

model, adjs = runDAZZLE(bl_data.X, DEFAULT_DAZZLE_CONFIGS)

get_metrics(model.get_adj(), bl_ground_truth)