Repository files navigation

This BNN BNI Analysis Tool can be used for predictive insight in Biological Networks through the use of Bayesian Neural Networks. This repository can be used for any biological datasets, however it is tailored toward individuals who are interested in neural networks/neural computation, computational biology, computational systems biology, computational genomics, computational pharmacology, and computational neuroscience. It is also made just as available to anyone who would like to use it for their own learning endeavours or to start their journey in any of these fields!

BNN_Biological_Network_Inference/├── data/│   └── example_data.csv│├── src/│   ├── data_preprocessing.py│   ├── train.py│   ├── inference.py│   └── utils.py│├── .gitignore├── LICENSE├── requirements.txt└── README.md

• Protein Data

• Genomic Data

• Drug Interaction Data

• Metabolite Data

• Microbiome Data

Your.csv files should follow this structure:

1. Features: Columns representing various features.
2. Target Variable: A column for the target variable that represents the interaction score or outcome you want to predict.

Example Structure:

feature1,feature2,feature3,target

value1,value2,value3,target_value

• Copy The Repository:

git clone https://github.com/peterajhgraham/BNN_Biological_Network_Inference.gitcd BNN_Biological_Network_Inference

• Format Your Data: Ensure your.csv files follow the structure outlined above.

• Place Data Files: Save your.csv files in thedata/ directory of the repository.

Install the required Python packages listed inrequirements.txt:

pip install -r requirements.txt

Run the data preprocessing script to load and prepare your data for modeling:

fromsrc.data_preprocessingimportload_data,preprocess_datafile_paths= {'example':'data/example_data.csv'}data=load_data(file_paths)processed_data=preprocess_data(data)

Use the training script to build and train your Bayesian Neural Network model with torchbnn:

fromsrc.trainimporttrain_modelinput_dim=processed_data['example'][0].shape[1]# Example for one datasetmodel=train_model(*processed_data['example'],input_dim)

After training, use the inference script to make predictions with your model:

fromsrc.inferenceimportinferpredictions=infer(model,processed_data['example'][1])

You can use the utility functions to visualize your results:

fromsrc.utilsimportplot_predictionsplot_predictions(processed_data['example'][2],predictions)