MscProjectDataAnalysis

This repository contains tools and scripts designed for analyzing and modeling biological data, with specific components for graph-based models and neural networks. It includes shell scripts, Python and R files, and Jupyter notebooks that facilitate the setup, analysis, and visualization of different data-driven models.

Repository Structure

• Shell Scripts (makeGraph.sh, makeMetaCells.sh, runGraphModel.sh, runNeuralNetwork.sh, etc.)

  • Contains bash scripts to automate various stages of data processing and model execution.

• Python Scripts (in src directory)

  • dataAnalysisPipeline.ipynb: Jupyter notebook for the primary data analysis pipeline.
  • graphModelFunctions.py, graphModels.py: Modules that define functions and implementations for graph-based modeling.
  • neuralNetworkFunctions.py, neuralNetworks.py: Modules for creating and training neural network models.
  • makeTFAPlots.py, plotLosses.py: Scripts for plotting and visualizing model performance metrics.

• R Scripts (in src and src/archive directories)

  • getVariableGenes.r, makeH5adFiles.r, mapToEnsembl.r: R scripts that assist in data preparation and transformation for analysis.
  • makeMetaCells.r, makeGraph.py: Scripts focused on preparing graph data and metacells for further analysis.

• Notebooks and Tutorials (in src and src/archive)

  • ATACseq.ipynb, DataAnalysisPipeline.ipynb, and others in the archive directory offer preliminary and additional analysis workflows.
  • Tutorials include CellOracle GRN models.ipynb, which demonstrates how to build gene regulatory network models using CellOracle.

Getting Started

1. Dependencies: Ensure all required Python and R libraries are installed. Refer to pythonRequirements.txt and requirements.R if available, or manually install based on script imports.
2. Data Preparation: Use the scripts in src to preprocess and prepare data. For example, makeH5adFiles.r prepares files in .h5ad format.
3. Run Models: Execute the models using provided shell scripts:
   • runGraphModel.sh to initiate graph-based models.
   • runNeuralNetwork.sh or other neural network scripts for deep learning models.
4. Visualization: Generate plots and performance metrics using scripts like makeTFAPlots.py and plotLosses.py.