proteoDeconv is an R package that facilitates deconvolution of bulk sample proteomic data to estimate proportions of cell types.
You can install the development version of proteoDeconv from GitHub with:
# install.packages("devtools")
devtools::install_github("ComputationalProteomics/proteoDeconv")The following example demonstrates a basic workflow using proteoDeconv. The input file report.pg_matrix.tsv in this case is from DIA-NN, and the LM7c.txt signature matrix was downloaded from the Decomprolute repository.
library(proteoDeconv)
pg <-
read_tsv("report.pg_matrix.tsv") |>
update_gene_symbols() |>
handle_gene_groups() |>
handle_missing_values(imputation_mode = "lowest_value") |>
handle_duplicate_genes(duplicate_mode = "slice") |>
handle_scaling(unlog = FALSE, tpm = TRUE)
signature <- read_tsv("LM7c.txt")
pg_deconvoluted <- deconvolute("cibersortx", pg, signature)To use proteoDeconv with CIBERSORTx, you will need to register on the CIBERSORTx website and request a token. This token is required for accessing the CIBERSORTx functionality and is free for academic users. The token and email used can be stored as environment variables in ~/.Renviron:
CIBERSORTX_TOKEN = your_token_here
CIBERSORTX_EMAIL = your_email_here
Additionally, Docker needs to be installed on the system in order to run CIBERSORTx.
For running proteoDeconv with the regular CIBERSORT (not CIBERSORTx), you will need to download the CIBERSORT source code from the CIBERSORTx website.
Additionally, you can retrieve ready-made signature matrices for cell type deconvolution from the Decomprolute repository on GitHub. These matrices can be used as inputs in the deconvolution process with proteoDeconv.