In combiz/scFlow: A Single-Cell/Nuclei Analysis Toolkit

A total of r prettyNum(sum(metadata$qc_summary$qc_cells_n_cells_passed, metadata$qc_summary$qc_cells_n_cells_failed), big.mark = ",") barcodes were submitted to the ambient RNA / empty droplet identification method emptyDrops. This identified r prettyNum(metadata$emptydrops_params$cells_found, big.mark = ",") cells and r prettyNum(metadata$emptydrops_params$emptydrops_found, big.mark = ",") empty droplets with an FDR cutoff of ≤ r sprintf("%1.2f%%", metadata$emptydrops_params$alpha_cutoff * 100). r if(!(toupper(metadata$emptydrops_params$retain_param) == "AUTO") & !is.null(metadata$emptydrops_params$retain_param)) {sprintf("The retain parameter was set to %s, therefore all barcodes with at least %s counts were assumed to contain cells.", metadata$emptydrops_params$retain, metadata$emptydrops_params$retain)} r if(is.null(metadata$emptydrops_params$retain)) {sprintf("The retain parameter was set to NULL, therefore the retain parameter was set to the total counts at the barcode rank knee point (%s) (i.e. all barcodes with at least %s counts were assumed to contain cells).", metadata$emptydrops_params$retain, metadata$emptydrops_params$retain)} r if(toupper(metadata$emptydrops_params$retain_param) == "AUTO") {sprintf("The retain parameter was set to 'auto', therefore the retain parameter was calculated as m/10 (%s) where m is the 99th percentile of the top %s (expect_cells parameter) barcodes by total UMI counts (i.e. all barcodes with greater than %s counts were retained as cells).", metadata$emptydrops_params$retain, metadata$emptydrops_params$expect_cells, metadata$emptydrops_params$retain)}

wzxhzdk:0

Monte-Carlo Iterations

The EmptyDrops algorithm was run with the specified r prettyNum(metadata$emptydrops_params$niters, big.mark = ",") Monte-Carlo iterations for the calculation of p-values. r if(metadata$emptydrops_params$n_limited > 0){sprintf("It was found that the p-values of %s non-significant cells were limited by the number of iterations, suggesting that a higher number of iterations would improve the cell / empty droplet calls.", metadata$emptydrops_params$n_limited)} else {sprintf("No barcodes were observed to have p-values limited by the number of iterations, suggesting that %s iterations was suitable for cell / empty droplet calling on this data.", prettyNum(metadata$emptydrops_params$niters, big.mark = ","))}

Ambient Profile

To evaluate whether the model was reliable, the p-values for all presumed ambient barcodes (i.e. total_counts <r metadata$emptydrops_params$lower) were extracted. Under the null hypothesis, the p-values for these barcodes should be uniformly distributed.

knitr::opts_chunk$set(echo = FALSE)
#ggplotly(metadata$emptydrops_plots$emptydrops_hist)
metadata$emptydrops_plots$emptydrops_hist

In addition to this visualization, the uniformity was tested using the r metadata$emptydrops_params$uniformity_method which yielded a p-value of r metadata$emptydrops_params$uniformity_pval (≤0.05 would suggest non-uniformity).

combiz/scFlow documentation built on Feb. 25, 2024, 10:25 a.m.