/Liver_Deconvolution

These are the scripts used for my paper on liver deconvolution using single-cell RNA-seq data

Primary LanguageR

Liver_Deconvolution

These are the scripts used for my paper on liver deconvolution using single-cell RNA-seq data

This manuscripts consists of three parts:

  1. Combining scRNA-seq data from healthy liver with scRNA-seq data from PBMCs.
  2. Deconvolution of bulk RNA-seq data from chronic liver disease patients
  3. WGCNA of bulk-RNA-seq data with clinical data and deconvolution results

Part 1: Combining scRNA-seq data from healthy liver with scRNA-seq data from PBMCs.

Data was obtained from:

scRNA-seq of five Healthy Livers - MacParland et al 2018 - GSE115469 These data were obtained from GSE as a raw count matrix .csv

scRNA-seq of 4 Healhy and 4 AH PBMCs - Kim et al 2020 - PRJNA596980 These data were generated by our lab, and analyzed as previously described See git repository - atomadam2/PBMC_AH_LPS_scRNA-seq for details and code for alignment using CellRanger After alignment, all data was uploaded into R using Seurat

LiverDeconv_scCombine.R - Script with code for pulling in all scRNA-seq data and combining This script will create an R object containing all data. This RObj is needed for all other scripts. Note: This will contain all PBMC data we generated from these samples, including cells treted with LPS. Subsequent analyses will remove this data for simplicity. The output of this script is an RObj used for all other scripts and analyses and can be used for future analyses

LiverDeconv_scFigures.R - Script with code for making all figures describing scRNA-seq data All parts of Figure 1 from paper are generated in this script (Except DE genes) Figure 2A Clustering is also made All Supplemental figures are made

LiverDeconv_scDE - Script with code for performing Resident vs Peripheral Differnetial Expression testing in Seurat This produces a large nnumber of tables for differentially expressed genes in different cell types Healthy Liver vs Healthy PBMC Healthy PBMC vs AH PBMC AH PBMC vs Healthy Liver Does not include liver specific cells (Hep, Chol, LSEC, Plasma cells) These tables can be put into other pathway analysis software (Metascape for example)

Kim2021_CibersortSigGene_SuppTable2 - This is Supplementary Table 2 in the manuscript, but as a text file. The original publication was converted to excel and has gene->date errors (sorry about that and thanks Guo-Liang Chew for finding this!)

Part 2: Deconvolution of bulk RNA-seq data from chronic liver disease patients

Reduce scRNA-seq data for deconvolution

LiverDeconv_Reduction.R - Script with code reducing scRNA-seq data Data are reduced by: Removing LPS treated cells Creating single separate clusters for CD4 T-cells, CD8 T-cells, NK-cells Removing clusters with very small numbers of cells (as described in paper) Subsampling all clusters to just 200 cells/cluster. This output can be put into Cibersortx for a signature gene matrix (Which is Supplemental Table 2)

LiverDeconv_CibersortBoxplots.R - Script with code for making boxplots of Cibersort data Simple script using ggplot to make clustered boxplots of cibersort data. Makes Figure 2B,C,D This script also requires the following files which contain all of the output from Cibersortx (basically all results)

  • Cibersort_CellTypes_Liver_numbers.csv
  • Cibersort_CellTypes_Liver.csv

Part 3: Analysis of Liver RNA-seq data

Pitt_AH_Liver_align.sh - This is a BASH script that very simply aligns the RNA-seq data to the human genome NOTES:

LiverDeconv_BulkRNASleuth.R - Script with code for running Sleuth to analyze RNA-seq data and making figure Makes all figiures for Figure 3 This script uses the following files:

  • Compiled_Names_v3.txt
  • Pitt_AH_TPM.txt

Part 4: WGCNA

LiverDeconv_WGCNA.R - Script with code to run WGCNA This script uses the following files:

  • Compiled_Names_v3.txt
  • Pitt_AH_TPM.txt