Skip to content

sashielvagus/Multimodal-Deconvolution-of-Zebrafish-Gene-Expression-Using-MuSiC-Project

Repository files navigation

Zebrafish Gene Expression Deconvolution Using MuSiC

This repository contains R scripts and resources for deconvolving bulk RNA-seq gene expression data in zebrafish, using MuSiC (Multi-subject Single-cell deconvolution). The goal is to investigate how environmental exposures influence gene expression at the cell-type level, using single-cell RNA-seq references from both pancreas-specific and whole-zebrafish sources.


Project Objective

To estimate cell type proportions in bulk RNA-seq samples of zebrafish and assess how they shift in response to environmental exposures. This is achieved through multimodal analysis by integrating:

  • Bulk RNA-seq gene counts (exposed vs control)
  • Single-cell RNA-seq reference data (annotated by cell type)

Script Details

deconvolution_bulk_zebrafish_to_zebrafish_pancreas.R

  • Uses pancreas-specific cell types (e.g., alpha, beta_1, duct)
  • Converts Seurat object (Control_annotated.rds) into a SingleCellExperiment
  • Performs MuSiC deconvolution and outputs:
    • Jitter plot comparing MuSiC vs NNLS estimates
    • Heatmap of estimated proportions
    • Stacked bar plot of cell type distributions
  • Reference Dataset:
    Control_annotated.rds downloaded from the Broad Institute's Single Cell Portal (SCP1549)

    Citation: Singh, S. P., Janjuha, S., Hartmann, T., Kayisoglu, O., Konantz, J., Birke, S., ... & Lickert, H. (2022). A single-cell atlas of de novo β-cell regeneration reveals the contribution of hybrid β/δ cells to diabetes recovery in zebrafish. Development, 149(2), dev199853. https://doi.org/10.1242/dev.199853


deconvolution_bulk_zebrafish_to_whole_zebrafish.R

  • Uses DanioCell 2023 dataset as the scRNA-seq reference (Daniocell2023_SeuratV4.rds)
  • Converts both bulk and single-cell data into ExpressionSet
  • Performs MuSiC deconvolution using the entire cellular atlas
  • Outputs the same set of visualizations and CSV matrices
  • Reference Dataset:
    Daniocell2023_SeuratV4.rds downloaded from the DanioCell Portal

    Citation: White, D. T., Freudenberg, J., Riva, A., Snyder, D. T., Lai, S. L., Ang, K. C., ... & Burgess, S. M. (2023). A zebrafish single-cell transcriptome atlas reveals shared and distinct cellular responses to injury. bioRxiv. https://doi.org/10.1101/2023.03.29.534734


Requirements

  • R (version ≥ 4.1)
  • CRAN & Bioconductor packages:
install.packages(c("ggplot2", "dplyr", "reshape2", "pheatmap"))
BiocManager::install(c("MuSiC", "Seurat", "SingleCellExperiment", "Matrix", "Biobase", "scuttle"))

License

This repository is shared under an academic research license. For permission to reuse or cite the work, please contact the author(s).

About

Multimodal project: MuSiC used to process gene expression data using a signature matrix built from single-cell RNA-seq data that defines gene profiles for each cell type. The goal of this project is to use multimodal to quantify how environmental exposures alter gene expression in zebrafish.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages