DetectCellStructuresBasedOnCellType: DetectCellStructuresBasedOnCellType

In bimberlabinternal/CellMembrane: A package with high-level wrappers and pipelines for single-cell RNA-seq tools

DetectCellStructuresBasedOnCellType

Description

A DBSCAN-based spatial cluster detection method to find dense cellular structures within spatial data. This function is currently implemented with FOV-based, iterative, classification in mind, but is technically extendable to global analysis of a single frame.

Usage

DetectCellStructuresBasedOnCellType(
  seuratObjectMetadata,
  cellTypeField = "cell_type",
  minimumClusterSizeCoefficient = 0.05,
  fovField = "fov",
  fovWhitelist = NULL,
  cellTypeWhiteList = NULL,
  xCoordinateField = "x_FOV_px",
  yCoordinateField = "y_FOV_px",
  substructureMetaDataFieldName = "Substructure",
  summarizeLocalResults = TRUE
)

Arguments

seuratObjectMetadata	Metadata dataframe storing, at minimum, a cell type or "transcriptomic clustering" type field, a field denoting FOV, and two spatial coordinates.
cellTypeField	A character field storing cell type annotations, but could be any discrete cell clustering assignment.
minimumClusterSizeCoefficient	The percentage of the input data (after cell type splitting) to be considered as a feasible minimum cluster size. Determines how many "noise data points" will be added by DBSCAN.
fovField	the metadata column that stores the Field of View information.
fovWhitelist	An optional whitelist of FOVs. By default, the function will loop over all FOVs, which could be time consuming.
cellTypeWhiteList	A vector of genes that constitute your substructure (e.g. c("Bcell", "BCell") for B cell follicles).
xCoordinateField	The metadata column that stores the x coordinate information within the Field of View.
yCoordinateField	The metadata column that stores the y coordinate information within the Field of View.
substructureMetaDataFieldName	An annotation that will be concatenated during the results. "Local" FOV information will be concatenated using "fov + substructureMetaDataFieldName + a substructure index" within the columns of the metadata.
summarizeLocalResults	An optional boolean that will wrap up the various substructureMetaDataFieldName columns into two single columns. One, which determines if a cell is within ANY of the defined substructures, stored in the output column "Within_Local + substructureMetaDataFieldName". The second is a metadata column that displays which of the local substructures the cell belongs in, concatenated as "Local + substructureMetaDataFieldName". "Local + substructureMetaDataFieldName + 0" is always the noise designation.

Value

Returns a dataframe containing columns related to the substructures found within the images at varying scopes. With summarizeLocalResults = FALSE, (number of FOVs) x (number of subtructures + 1) columns will be added. summarizeLocalResults rolls these high resolution results into two additional columns relative to the fovField.

Examples

## Not run: 
#Perform Cell Structure detection for B cell follicles. 

metadata <- DetectCellStructuresBasedOnCellType(seuratObjectMetadata, 
cellTypeField = "cell_type", 
minimumClusterSizeCoefficient = 0.05,
fovField = "fov",
fovWhitelist = 1,
cellTypeWhiteList = c("Bcell", "B_cell", "B.cell"),
xCoordinateField = "x_FOV_px", 
yCoordinateField = "y_FOV_px", 
substructureMetaDataFieldName = "BCF",
summarizeLocalResults = TRUE
)

#load/install packages for plotting
library(pacman)
p_load(ggplot2, dplyr, egg, patchwork)

#define plotting layout
layout <- "
#AAAA#
BBBCCC
"

#Plot results
ggplot(metadata %>% filter(fov == 1), 
aes(x = x_FOV_px, y = y_FOV_px, color = simple_cellType)) + 
geom_point() + 
egg::theme_article() + 
ggtitle('Cell Type assignment')
ggplot(metadata %>% filter(fov == 1), aes(x = x_FOV_px, y = y_FOV_px, color = factor(Local_BCF))) + 
geom_point() + 
egg::theme_article() +  
ggtitle('Specific substructure cell assignment') + 
ggplot(metadata %>% filter(fov == 1), 
aes(x = x_FOV_px, y = y_FOV_px, color = Within_Local_BCF)) + 
geom_point() + 
egg::theme_article() + 
ggtitle('Non-specific substructure cell assignment') + 
plot_layout(design = layout, guides = "collect")

## End(Not run)

bimberlabinternal/CellMembrane documentation built on Oct. 16, 2024, 6:53 a.m.