izar_transform: generalizes the transformation from Izar et al 2020 for...

In trichelab/velocessor: The world's dumbest single-cell RNA velocity calculator and loader

Description

In order to compare the results of plate-seq and droplet-seq preparations, Izar and coauthors proposed a transformation of transcripts per million (TPM) for transcript i in cell j from a non-UMI plate-seq library as:

Usage

izar_transform(txis, orig = "tpm", dupes = 10, pseudo = 1)

Arguments

txis	a SingleCellExperiment
orig	the name of the assay to transform ("tpm")
dupes	an estimated duplication rate (default is 10, per Izar)
pseudo	a pseudocount to add to (TPM/dupes) (default 1, ibid)

Details

Ei,j = log2( (TPMi,j / 10) + 1 )

Here we generalize this slightly by allowing a variable scaling and offset. The goal is to make non-UMI plate-seq and UMI droplet-seq data comparable. We caution the user that 'comparable' is a subjective term here.

Most plate-seq protocols do not add unique molecular indices (UMIs) to each template fragment in a library (plate-seq protocols WITH UMIs include Quartz-Seq2, SMART-Seq2, and STORM-UMI). In contrast, almost all droplet protocols apply both a cell barcode (CB) and UMI barcode (UB) to each template fragment, disambiguating whether two fragments that map to the same reference sequence are from the same template molecule or not. In order to compare the results of plate-seq and droplet-seq preparations, Izar and colleagues proposed this transformation. One may additionally apply methods such as sctransform to the resulting estimate, harmonize the resulting matrix, or similar shenanigans. Alternatively, one may use a UMI-enabled plate-seq preparation to elide this transformation.

Value

       a SingleCellExperiment with assay `izar`

References

Izar B, Tirosh I, Stover EH, et al. A single-cell landscape of high-grade serous ovarian cancer. Nat Med 26, 1271–1279 (2020). https://doi.org/10.1038/s41591-020-0926-0

trichelab/velocessor documentation built on Jan. 5, 2022, 6:27 p.m.