RISE Package

RISE is an R package designed for tackling the zero inflation problem in sequence-based spatial transcriptomic (ST) data by leveraging neighborhood-based regression models.

Installation

You can install RISE directly from Github

devtools::install_github("YSPeng1225/RISE")
library(RISE)

or you can first download RISE and install it manually in your local computer

install.packages("path/to/RISE", repos = NULL, type = "source")
library(RISE)

Example Data

To help users understand the data structure, RISE includes an example dataset, spot_example, with two elements:

• spot_express: a sparse matrix of gene expression values with genes as rows and spots as columns.

• coordinates: a matrix containing spatial x and y coordinates for spots.

Load and print the example data:

data(spot_example)
print(spot_example$spot_express[1:10,1:10])
print(spot_example$coordinates[1:10,])
# normalization
norm_data <- log2(t(t(spot_example$spot_express)/colSums(spot_example$spot_express) * median(colSums(spot_example$spot_express))) + 1)

RISE Usage

RISE mainly relies on two functions, FindNeighbors and RISE, with the previous focusing on the neighborhood selection and the latter aiming to imputing zeros using regression models.

In FindNeighbors function, users can adjust the neighbor number of spots through the parameter thresh (range from 0 to 1), with larger thresh including more neighbors. FindNeighbors returns a Boolean matrix containing the neighbors' indicators, which facilitates the direct application in RISE.

# neighbors identification
neighbors <- FindNeighbors(spot_example$coordinates, thresh = 0.3)

# to summarize the number of neighbors
summary(rowSums(neighbors))
hist(rowSums(neighbors), breaks = 200)

# to obtain the neighbors' names
spots <- colnames(neighbors)
neighbor_names <- apply(neighbors, 1, function(x) spots[x])
neighbor_names[["spot29"]]

Once the value of thresh is determined by the user, the RISE function can be used to obtain the imputed gene expression matrix.

spot_imputed <- RISE(spot_example$spot_express, spot_example$coordinates, thresh = 0.3)
# after imputation
spot_imputed[1:5,1:5]
# before imputation
norm_data[1:5,1:5]