process_velo_txis: Load some Alevin runs quantified with velocity information

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

Description

This is evolving rapidly to handle microwell-seq and plate-seq data.

Usage

process_velo_txis(
  runs,
  txstub,
  qm = "alevin/quants_mat.gz",
  QC = TRUE,
  HARMONY = FALSE,
  CLUSTER = TRUE,
  DEDUPE = FALSE,
  SCVELO = FALSE,
  meta = FALSE,
  ...,
  BPPARAM = SerialParam()
)

Arguments

runs	a named vector of paths (if unnamed, you'll have problems)
txstub	the shared portion of the transcriptome annotation files
qm	the name and path of the Alevin quants matrix file in 'runs'
QC	do rudimentary quality control on each Alevin dataset? (TRUE)
HARMONY	run Harmony on merged dataset? (FALSE, as harmony is non-BioC)
CLUSTER	apply simple Louvain clustering to the result? (TRUE)
DEDUPE	de-dupe genes and cells (doublets)? (FALSE unless SCVELO=TRUE)
SCVELO	run scVelo on the (perhaps harmonized) result? (FALSE, slow)
meta	use tximeta for import_velo_txis? (FALSE; avoid bfc issues)
...	other arguments to pass to velociraptor::scvelo, if any
BPPARAM	if running in parallel, provide a MulticoreParam or similar

Details

The results of this function can be further QC'ed and/or batch corrected, for example with Harmony or sva::ComBat, then passed to velociraptor::scvelo and/or velocessor::cellrank for data-driven cell fate inference. We are investigating various factorization approaches that may allow for sensible batch effect correction while respecting the relationships between spliced and unspliced transcript abundance, but have not yet settled on a default.

See also plot_velo for a relatively straightforward exploratory plot based on UMAP coordinates and the downsampled subset of cells from the SingleCellExperiment that this function and velociraptor::scvelo yields.

Value

   a SingleCellExperiment with 

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