This document provides details on how to use the functions for generating various types of simulated spatial transcriptomics data, including descriptions of the specific code and parameter meanings for each data type, as well as instructions on generating simulated SRT data from custom scRNA-seq and STARmap data.
SSTD is based on R and can be easily installed on Windows, Linux as well as MAC OS.
First, users should install R >= 4.1.0.
Next, install devtools and SSTD.
Make sure you have R and the necessary packages installed. The key packages used in these functions include 'ggplot2' and 'scatterpie'.
install.packages(c("ggplot2", "scatterpie"))
#load the packages
library(ggplots)
library(scatterpie) if (!requireNamespace("devtools", quietly = TRUE)) {
install.packages("devtools")
}
devtools::install_github("YourUsername/SSTD")
#load the SSTD package
library(SSTD) This function processes the input data files, generates synthetic spatial spot true proportions and locations, creates gradation arrays for specified cell types, calculates nUMI counts, and generates and saves plots.
.
- 'meta_data_file': Path to the metadata file (CSV format).
- 'gene_count_file': Path to the gene count file (CSV format).
- 'selected_clusters': A named vector specifying the clusters to include (e.g., c(celltype1="eL4", celltype2="eL5", celltype3="eL6", celltype4="Endo")).
- 'prefix': Prefix for the output CSV files.
- 'NULL': The function performs operations and saves results to files but does not return a value.
prefix = "type_one"
source("R/utilities.R")
meta_data_file <- system.file(package="SSTD", "extdata", "meta_data.csv")
gene_count_file <- system.file(package="SSTD", "extdata", "gene_count.csv")
selected_clusters <- c(celltype1="eL4", celltype2="eL5", celltype3="eL6", celltype4="Endo")
generate_type1_data(meta_data_file, gene_count_file, selected_clusters, "type_one") Ensure the metadata and gene count data are in CSV format and specify the file paths.
Use the generate_type1_data function to process the data and generate results.
generate_type1_data(meta_data_file, gene_count_file, selected_clusters, prefix) - 'org_data_space' reads the metadata and gene count files, filters data based on selected clusters, and saves the processed data.
org_data <- org_data_space(meta_data_file, gene_count_file, output_csv_file = paste0(prefix, "_org_data_space.csv"), selected_clusters) - 'spatial_spot_true_prop' creates spatial spot true proportions data for the selected clusters and saves it.
sim_spatial_spot_true_prop_data <- spatial_spot_true_prop(selected_clusters, filename = paste0(prefix, "_spatial_spot_true_prop.csv")) - 'spatial_spot_loc' generates spatial spot locations and saves the data.
spatial_spot_loc(filename = paste0(prefix, "_spatial_spot_loc.csv")) - 'get_gradation_2d' generates 2D gradation arrays for each cell type.
# celltype1
array <- get_gradation_2d(1, 1, 10, 20, TRUE)
array_1 <- get_gradation_2d(0, 0, 30, 20, TRUE)
get_gradation_2d_celltype1 <- rbind(array, array_1)
# celltype2
array <- get_gradation_2d(0, 0, 10, 20, TRUE)
array_0 <- get_gradation_2d(1, 1, 10, 20, TRUE)
array_1 <- get_gradation_2d(0, 0, 20, 20, TRUE)
get_gradation_2d_celltype2 <- rbind(array, array_0)
get_gradation_2d_celltype2 <- rbind(get_gradation_2d_celltype2, array_1)
# celltype3
array <- get_gradation_2d(0, 0, 10, 20, TRUE)
array_0 <- get_gradation_2d(1, 1, 10, 20, TRUE)
array_1 <- get_gradation_2d(0, 0, 20, 20, TRUE)
get_gradation_2d_celltype3 <- rbind(array_1, array_0)
get_gradation_2d_celltype3 <- rbind(get_gradation_2d_celltype3, array)
# celltype4
array <- get_gradation_2d(1, 1, 10, 20, TRUE)
array_1 <- get_gradation_2d(0, 0, 30, 20, TRUE)
get_gradation_2d_celltype4 <- rbind(array_1, array) - 'spatial_spot_nUMI' calculates the total nUMI counts for all cell types and saves the results.
result_matrices <- list(result_matrix(get_gradation_2d_celltype1),
result_matrix(get_gradation_2d_celltype2),
result_matrix(get_gradation_2d_celltype3),
result_matrix(get_gradation_2d_celltype4))
new_data <- spatial_spot_nUMI(sim_spatial_spot_true_prop_data, result_matrices, selected_clusters, org_data, prefix) - 'generate_plots_and_save' generates and saves plots of cell locations and proportions.
combined_arrays <- list(selected_clusters, get_gradation_2d_celltype1,
get_gradation_2d_celltype2,
get_gradation_2d_celltype3,
get_gradation_2d_celltype4)
generate_plots_and_save(combined_arrays, selected_clusters, new_data, sim_spatial_spot_true_prop_data, prefix)The generate_type2_data function simulates spatial transcriptomics data, which is essential for studying gene expression within a spatial context. The function performs the following steps:
.
Reads and filters the metadata and gene count files based on the selected clusters.
Simulates the true proportions of different cell types in spatial spots.
Generates spatial coordinates for each spot.
Creates 2D gradation patterns to model the spatial distribution of each cell type.
Computes the number of unique molecular identifiers (nUMIs) for each spot, reflecting the abundance of gene transcripts.
Creates visual representations of the spatial distribution and proportions of cell types.
The output includes several CSV files containing the simulated data and plots illustrating the spatial patterns of cell types and gene expression.
.
The generate_type3_data function simulates spatial transcriptomics data for selected clusters with a distinct gradation pattern compared to other types. This function is designed to capture complex spatial relationships in gene expression data.
.
The generate_type3_data function simulates spatial transcriptomics data for selected clusters with a distinct gradation pattern compared to other types. This function is designed to capture complex spatial relationships in gene expression data.
.
This guide provides a step-by-step process for generating simulated spatial transcriptomics (SRT) data using custom single-cell RNA sequencing (scRNA-seq) data and spatial transcriptomics data. The simulation is performed using the Generating_simulation_data function from the R programming environment.
Prepare your data in the following formats:
- Spot Spatial Count: A matrix of gene expression counts for each cell, similar to the example provided in the function documentation.
- Simulated Spatial Spot Location: A table with the spatial coordinates (X and Y) for each spot.
- Cell Cell Type: A table with the cell type information, including cluster IDs and cell type names.
Ensure that your data files have the correct column names as specified in the function parameters.
spot_spatial_count_path <- "path/to/your/spot_spatial_count.txt"
sim_spatial_spot_loc_path <- "path/to/your/sim_spatial_spot_loc.txt"
cell_celltype_path <- "path/to/your/cell_celltype.txt" grid_interval <- 500 # Adjust based on your spatial resolution
prefix <- "my_simulation" # Prefix for the output files Generating_simulation_data(
spot_spatial_count = spot_spatial_count_path,
sim_spatial_spot_loc = sim_spatial_spot_loc_path,
cell_celltype = cell_celltype_path,
grid_interval = grid_interval,
prefix = prefix
) 7.2.4 The function will generate several output files with the specified prefix, including simulated gene expression data, spatial locations, cell type proportions, and more.
- _sim_spatial_spot_nUMI.csv: Simulated gene expression data for pseudo-spots.
- _sim_spatial_spot_loc.csv: Physical locations of simulated pseudo-spots.
- _sim_spatial_spot_true_prop.csv: True cell type proportions of simulated pseudo-spots.
- _cells_nUMI.csv: Gene expression data of spot cells.
- _cells_celltype.csv: Cell type annotation of spot cells.
The function also generates visualizations such as:
- _plot_Locations_of_cell_with_grid.pdf: Plot showing the locations of cells with a grid overlay.
- _Pie_chart_of_cell_type_proportions.pdf: Pie chart representing cell type proportions.
- Ensure that your data files are correctly formatted and have the necessary column names.
- Adjust the grid_interval parameter based on the spatial resolution of your data.
- Customize the prefix to organize your output files as needed.