uflcod / miRNA-periodontal-disease

miRNA periodontal disease

Purpose

This repository is for developing methods to analyze the assocations between miRNAs and periodontal disease.

Notebooks

The repository contains the following notebooks in the notebooks directory.

AI Notebooks

• xgboost_miRNA.ipynb
  Runs XGBoostClassifer on three datasets:

  • Sham and infeceted mice from all weeks; i.e., 8 week and 16 week datasets are merged.
  • Sham and infected mice at 8 weeks.
  • Sham and infected mice at 16 weeks.

  In each dataset, there is a flag (named ‘infected’) that marks whether the mice came from the infected group or the sham group.
  XGBoost's variable importance and SHAP values are then used to determine which miRNA variable was most important in each cohort dataset.

Utility Notebooks

• transpose_merge_miRNA.ipynb
  Run notebook to transpose and merge the miRNA data files into a single transposed_Tf_miRNA.xlsx file.
• transpose_Tf_aveolar_bone_resporption.ipynb
  Run notebook to transpose the Tf_aveolar_bone_resporption.xlsx data into the transposed_Tf_aveolar_bone_resporption.xlsx file.
• merge_miRNA_bone_resorption.ipynb
  Run notebook to create the merged_miRNA_resporption.xlsx data file.

Virtual environment setup

• To create a virtual environment run the command python -m venv venv. This will create a virtual environment in the venv directory. You can specify a different directory name if you wish.
• Run source venv/bin/activate to activate (i.e., run) the virtual environment.
• After activating the environment, install the project libraries using the command pip install -r requirements.txt -U.

Jupyter hack for virtual environments

Jupyter notebooks won't necessarily have access to the libraries installed in the virtual environment.
One hack to get around this is to create a softlink to the Jupyter binary created in the virtual environment like so ln -s venv/bin/jupyter.

You can then start Jupyter within the virtual environment using the softlink. E.g., ./jupyter lab.
The allows the Jupyter notebook to access the libraries in the virtual environment.

Another option is create a Jupyter kernel from the virtual environment. See here for details.