Barcode ranks per sample

barcodeRanksPerSample(object, ...)

# S4 method for SingleCellExperiment
barcodeRanksPerSample(object, ...)





Passthrough arguments to DropletUtils::barcodeRanks().




Analyses of droplet-based scRNA-seq data often show a plot of the log-total count against the log-rank of each barcode, where the highest ranks have the largest totals. This is equivalent to a transposed empirical cumulative density plot with log-transformed axes, which focuses on the barcodes with the largest counts. The barcodeRanks function will compute these ranks for all barcodes. Barcodes with the same total count receive the same average rank to avoid problems with discrete runs of the same total.

The function will also identify a number of interesting points on the curve for downstream use, namely the inflection and knee points. Both of these points correspond to a sharp transition between two components of the total count distribution, presumably reflecting the difference between empty droplets with little RNA and cell-containing droplets with much more RNA.

  • The inflection point is computed as the point on the rank/total curve where the first derivative is minimized. The derivative is computed directly from all points on the curve with total counts greater than lower. This avoids issues with erratic behaviour of the curve at lower totals.

  • The knee point is defined as the point on the curve where the signed curvature is minimized. This requires calculation of the second derivative, which is much more sensitive to noise in the curve. To overcome this, a smooth spline is fitted to the log-total counts against the log-rank using the smooth.spline function. Derivatives are then calculated from the fitted spline using predict.

We supply a default setting of df to avoid overfitting the spline, as this results in unstability in the higher derivatives (and thus the curvature). df and other arguments to smooth.spline can be tuned if the estimated knee point is not at an appropriate location. We also restrict the fit to lie within the bounds defined by fit.bounds to focus on the region containing the knee point. This allows us to obtain an accurate fit with low df, rather than attempting to model the entire curve.

If fit.bounds is not specified, the lower bound is automatically set to the inflection point, which should lie after the knee point. The upper bound is set to the point at which the first derivative is closest to zero, i.e., the “plateau” region before the knee point. Note that only points with total counts above lower will be considered, regardless of how fit.bounds is defined.


Requires DropletUtils package to be installed.

Updated 2019-09-17.

See also


data(SingleCellExperiment, package = "acidtest") ## SingleCellExperiment ==== if (requireNamespace("DropletUtils", quietly = TRUE)) { object <- SingleCellExperiment x <- barcodeRanksPerSample(object) names(x) }
#> [1] "sample1" "sample2"