MuSiC implementation in Julia - Under Construction

This project allows for data preparation and analysis in the MuSiC (Multi-Sample Single-Cell) pipeline, which has been refactored into Julia for speed. This project is not complete yet and remains under construction.

Inspiration

This code was based on the following: LuoHan's Python implementation of MuSiC: https://github.com/Prayforhanluo/MuSiC_Python/blob/master/MuSiC.py

Bulk tissue cell type deconvolution with multi-subject single-cell expression reference: https://www.nature.com/articles/s41467-018-08023-x#Sec8

Raison d'etre

The original MuSiC implementation is done in R, and I wanted to see if implementing it in Julia yieleded performance benefits when used on larger datasets.

Project Structure

• main.jl: The main entry point of the application.
• src/: This directory contains the main modules of the application.
  • data_analysis.jl: Contains functions for data analysis.
  • data_loading.jl: Contains functions for loading data.
  • data_preparation.jl: Contains functions for preparing data.
• Project.toml and Manifest.toml: These files are used by Julia's package manager.

How to Run

To run this project, you need to pass 4 arguments to the main.jl script. These arguments are the paths to the files that contain the SC Meta Info, SC Count Info, bulk Meta Info, and bulk Count Info, respectively.

julia main.jl path_to_sc_meta_info path_to_sc_count_info path_to_bulk_meta_info path_to_bulk_count_info

Dependencies

This project uses the following Julia packages:

• DelimitedFiles
• DataFrames
• LinearAlgebra
• Statistics

The environment can be replicated by calling instantiate from Julia's environment manager.