The 10X document says the cell calling algorithm for expression data and TCR data are quite different . I want to make sure that if I can still use the DropletUtils to identify cells in 10X 5` TCR data. If can , how to perform , which functions shold I employ and which paramaters shouls I pay attentions.
I don't deal much with TCR data, so I can't really say for sure. But I don't see why you couldn't use emptyDrops. In principle, the same problem of ambient contamination would apply for TCR droplet data. One could imagine testing whether each droplet's TCR sequences were significantly different from the TCR sequences detected in the ambient pool.
Practically speaking, there are two major challenges. The first is wrestling the CellRanger TCR output data frame into a matrix that emptyDrops will accept. This would require basically one row for every unique sequence for every TCR component; possibly a lot of rows, which will make emptyDrops run a bit slower. The other problem is whether or not the defaults are appropriate - lower=100 is a pretty sensible default for RNA-seq, but you will have to judge this separately based on the UMI counts in TCR data. Similarly, if the barcode rank plots don't have a knee point, I would disable the retain= option.