Weighted least squares (WLS) in scMultiMap
scMultiMap_WLS.RdThis function implements the 2nd step of scMultiMap: estimate and infer peak-gene association with WLS. It takes the output from the 1st step, scMultiMap_IRLS as input.
Source
Cell-type-specific mapping of enhancer and target genes from single-cell multimodal data. Chang Su, Dongsoo Lee, Peng Jin, Jingfei Zhang; https://www.biorxiv.org/content/10.1101/2024.09.24.614814v1
Arguments
- count_list
A list of two p by n count matrices. The first element is for gene and the second for peak.
- seq_depth_list
A list of two length p vectors of sequencing depths. The first element is for scRNA-seq and the second for scATAC-seq.
- pairs_df
A data frame of candidate peak-gene pairs to evaluate. The first column is gene name, and the second is peak name.
- irls_list
A list of two lists generated from the 1st step of scMultiMap, i.e. scMultiMap_IRLS. The first list contains the results on genes, the second on peaks.
- verbose
Whether to print detailed messages. Default to FALSE.
Value
A data.frame with five columns
- gene
gene in the peak-gene pair
- peak
peak in the peak-gene pair
- pval
p-value
- test_stat
test statistics
- covar
estimated covariance
Each row corresponds to one peak-gene pair's results from scMultiMap. pval and test_stat denote the statistical significance of peak-gene association, and covar denote the magnitude of association. The number of rows is equal to the number of pairs in pairs_df present in the Seurat object.
Examples
count_list <- list(gene = t(as.matrix(small_obj[['RNA']]$counts)),
peak = t(as.matrix(small_obj[['peak']]$counts)))
seq_depth_list <- list(gene = Matrix::colSums(small_obj[['RNA']]$counts),
peak = Matrix::colSums(small_obj[['peak']]$counts))
wls_res <- scMultiMap_WLS(count_list, seq_depth_list, small_pairs_df, small_irls_list)
#> There are 10 unique genes in the peak-gene pairs.