spase: Main function for detecting spatial allele-specific...

In lulizou/spASE: Detecting Allele-Specific Expression in Spatial Transcriptomics

Main function for detecting spatial allele-specific expression (ASE) patterns

Description

This function fits a spatial beta-binomial model, estimates a 2D smoothed probability function, and stores information for plotting results. Uses a likelihood ratio test to test for a significant spatial fit over optional baseline covariates such as cell type.

Usage

spase(
  matrix1,
  matrix2,
  covariates,
  method = "betabinomial",
  df = 5,
  genes = NULL,
  min.pixels = 100,
  min.pixels.per.factor = 15,
  min.umi = 500,
  cores = 1,
  verbose = F
)

Arguments

matrix1	matrix or sparseMatrix of allele counts for allele to be modeled. Row names should be genes and column names should be pixel IDs.
matrix2	matrix or sparseMatrix of allele counts for second allele.
covariates	data frame that matches pixel IDs to spatial coordinates and other covariates. First column assumed to be pixel IDs, next two columns assumed to be x,y coordinates. Any additional columns assumed to be covariates that will be included in the baseline model.
method	string specifying which method to use. Default is 'betabinomial'. Other options are 'quasibinomial' and 'apeglm'.
df	integer, sets the number of degrees of freedom to use for the smoothing spline. Default is 5. Usually want to increase this or test multiple.
genes	optional vector of genes to run on. Good for testing things out.
min.pixels	integer specifying the minimum number of pixels a gene should be present on to fit. Default is 100.
min.pixels.per.factor	integer specifying the minimum number of pixels a gene should be present on if there are factor covariates e.g. cell type. For example, the default of 15 requires that at least 15 pixels are that cell type, otherwise that cell type will not be included in the model for that gene.
min.umi	integer specifying the minimum number of UMI a gene should have to fit. Default is 500.
cores	number of cores to use for parallelization. Default is 1.
verbose	whether or not to print a lot of status messages. Default is FALSE.

Details

Fits a beta-binomial thin plate regression spline for each gene.

Value

If method is betabinomial or quasibinomial, a list containing the following output:

• results a data frame containing a summary of the results including p-values and q-values for goodness of spatial fit over baseline covariates provided

• fits a list of beta-binomial model fit objects for each gene

.

lulizou/spASE documentation built on May 22, 2024, 5:24 a.m.