We developed D2Cell-pred, a hybrid model that combines mechanistic and deep learning approaches to predict outcomes for new cell factories. D2Cell-pred takes as input the target product, the GEM structure, and a set of gene modifications, and outputs the predicted impact of these modifications on the product.
We used the following Python packages for core development. We tested on Python 3.9.
| name | version |
|---|---|
| numpy | 1.24.4 |
| pandas | 2.0.3 |
| networkx | 3.1 |
| tqdm | 4.66.5 |
| torch | 2.4.0 |
| torch-geometric | 2.5.3 |
| scipy | 1.10.1 |
| seaborn | 0.13.2 |
| scikit-learn | 1.3.2 |
Clone codes and download necessary data files
- (1). Download the D2Cell-pred package
git clone https://github.com/LiLabTsinghua/D2Cell.git- (2). Download required Python package
pip install -r requirements.txt- (3). Download and unzip the model parameters under D2Cell
- (4). Run Code/D2Cell-pred Model/predict demo.ipynb demo
We also provide an dataset web server: D2Cell.
- Feiran Li (@feiranl), Tsinghua University, Shenzhen, China
- Xiongwen Li (@xiongwenL), Tsinghua University, Shenzhen, China