knee_plot: Plot the transposed knee plot and inflection point
In BUSpaRse: kallisto | bustools R utilities

Description Usage Arguments Value Note Examples

Plot a transposed knee plot, showing the inflection point and the number of remaining cells after inflection point filtering. It's transposed since it's more generalizable to multi-modal data.

get_knee_df(mat)

get_inflection(df, lower = 100)

knee_plot(df, inflection)

`mat`	Gene count matrix, a dgCMatrix.
`df`	The data frame from `get_knee_df`.
`lower`	Minimum total UMI counts for barcode for it to be considered when calculating the inflection point; this helps to avoid the noisy part of the curve for barcodes with very few counts.
`inflection`	Output of `get_inflection`.

get_knee_df returns a tibble with two columns: total for total UMI counts for each barcode, and rank for rank of the total counts, with number 1 for the barcode with the most counts.

get_inflection returns a numeric(1) for the total UMI count at the inflection point.

knee_plot returns a ggplot2 object.

Code in part adapted from barcodeRanks from DropetUtils.

# Download dataset already in BUS format
library(TENxBUSData)
TENxBUSData(".", dataset = "hgmm100")
tr2g <- transcript2gene(c("Homo sapiens", "Mus musculus"), 
  type = "vertebrate",
  ensembl_version = 99, kallisto_out_path = "./out_hgmm100")
m <- make_sparse_matrix("./out_hgmm100/output.sorted.txt",
  tr2g = tr2g, est_ncells = 1e5,
  est_ngenes = nrow(tr2g), TCC = FALSE)
df <- get_knee_df(m)
infl <- get_inflection(df)
knee_plot(df, infl)