README.md

In ajwilk/EpicTools:

EpicTools

NOTE: this repository is no longer actively maintained

Tools for pre-processing of count matrices for droplet-based scRNA-seq analysis

This package contains several functions that process scRNA-seq count matrices built by dropEst before initialization of a Seurat object for post-processing. 1. EpicPre filters and parses layered exonic, intronic, and spanning count matrices. 2. mergeCM rapidly and iteratively merges up to 48 sparse matrices in a single command.

Installation

devtools::install_github('ajwilk/EpicTools')

Usage

The output of dropEst -V (https://github.com/hms-dbmi/dropEst) is an .rds file that contains 3 dgCMatrix objects (exon/intron/spanning) for a given scRNA-seq sample.

First, read in the count matrix files you wish to pre-process:

path = "path_to_dropEst_output/"
cm.list = paste0(path, list.files(pattern = "*.matrices.rds", path = path))
cm.files <- lapply(cm.list, readRDS)
names(cm.files) <- sub("\\_cell.counts.matrices.rds", "", list.files(pattern = "*.matrices.rds", path = path))

The count matrices are now stored as a list that can be processed by EpicPre(), which filters count matrices built by dropEst by the number of UMIs per cell, complexity per cell (the number of genes divided by the number of UMIs), and the percentage of mitochondrial and rRNA (if applicable) reads. There are two EpicPre() functions that come with different defaults for filtering metrics. EpicPreHS() is designed for processing human scRNA-seq data, and EpicPreMN() for M. nemestrina data.

EpicPre() will also print several plots to help you QC each sample and ensure that the filtering thresholds are set appropriately. Defaults are shown below:

EpicPreHS(cm_name, min.counts = 1e3, max.counts = 15e3, max.complexity = 0.75, max.percent.mito = 0.2, max.percent.rRNA = 0.2)
EpicPreMN(cm_name, min.counts = 1e3, max.counts = 15e3, max.complexity = 0.75, max.percent.mito = 0.001)

This function parses and processes dropEst objects that must contain exonic, intronic, and spanning count matrices, created using the -V flag.

cm.pp <- mapply(EpicPreHS, cm.files, orig.ident = names(cm.files), SIMPLIFY = F)

Next, the filtered count matrices can be merged together using mergeCM. This function will accept up to 48 count matrices for merging.

combined.emat <- mergeCM(cm.pp, type = "emat")
combined.nmat <- mergeCM(cm.pp, type = "nmat")

combined.emat is now ready to be used as an input count matrix for Seurat. Cell names contain the name of each cm_file followed by "." followed by a number 1:ncol(cm_file) followed by "_" and the cell barcode.

ajwilk/EpicTools documentation built on April 1, 2022, 7:02 a.m.