In celsomilne/scSpatial: Single-Cell Spatial Data

library(BiocStyle)
knitr::opts_chunk$set(error=FALSE, message=FALSE, warning=FALSE)

Downloading the Data

First, we download the data from Mendeley.

library(BiocFileCache)
bfc <- BiocFileCache("raw_data", ask=FALSE)

cellData.url <- "https://data.mendeley.com/datasets/cydmwsfztj/2/files/4473bd0c-b617-4c79-8253-8d61fbe4e8e8/Cells.zip?dl=1"
cellTypes.url <- "https://data.mendeley.com/datasets/cydmwsfztj/2/files/59e8da72-5bfe-4289-b95b-28348a6e1222/CellTypes.zip?dl=1"
metadata.url <- "https://data.mendeley.com/datasets/cydmwsfztj/2/files/b0405cb9-2816-4bc9-9baa-16d0f509d873/Metadata.csv?dl=1"
all.urls <- c(cellData.url, cellTypes.url, metadata.url)

data.path <- bfcrpath(bfc, all.urls)

Now read it into our session.

library(tidyverse)

temp <- file.path("temp/")
dir.create(temp, recursive = TRUE)
lapply(data.path, FUN=unzip, exdir=temp)

# Read in datasets.
allCells <- read_csv(file.path(temp, "All_Cells.csv"))
celltype <- read_csv(file.path(temp, "CellTypes.csv"))
meta <- read_csv(file.path(data.path[["BFC3"]]))

Now, we want to clean the data up a bit.

# Load libraries
library(tidyverse)
library(spicyR)

# Clean celltype
celltype <- celltype %>%
  mutate(imageID = core, ImageNumber = as.numeric(as.factor(celltype$core)), imageCellID = id, ObjectNumber = as.numeric(lapply(strsplit(id,"_"),function(x)x[2])), cellType = factor(CellType))

cells <- allCells %>%
  mutate(x = AreaShape_Center_X, y = AreaShape_Center_Y) %>%
  select(x,y,ImageNumber, ObjectNumber, starts_with("Intensity_MeanIntensity_CleanStack_"), starts_with("AreaShape_")) %>%
  inner_join(celltype, by = c("ImageNumber", "ObjectNumber"))

cellExp <- SegmentedCells(as.data.frame(cells), 
                          intensityString = "Intensity_MeanIntensity_CleanStack_", 
                          morphologyString = "AreaShape_")

meta$stage <- factor(as.character(meta$stage),levels = c("Non-diabetic", "Onset", "Long-duration"))
meta <- meta %>% mutate(imageID = image) %>% select(-image)
colnames(meta) <- paste0("phenotype_", colnames(meta))
meta <- meta %>%
  rename(imageID=phenotype_imageID)

data <- as.data.frame(cellExp)
data <- data %>% 
  left_join(meta, by="imageID")

Save to ExperimentHub Storage

Prepare all the files for storage on the ExperimentHub.

path <- file.path("scSpatial", "Damond-Diabetes", "1.0")
dir.create(path, showWarnings=FALSE, recursive=TRUE)

saveRDS(data, file=file.path(path, "spatialCellData.rds"))

And get rid of the temporary folder and the cache.

unlink(temp, recursive = TRUE, force = TRUE)
unlink("raw_data", recursive = TRUE, force = TRUE)

celsomilne/scSpatial documentation built on Aug. 11, 2020, 9:51 a.m.