Question: Is there an obvious way of telling which barcodes contain multiple cells (multiplets) for V(D)J libraries?
Answer: At this time we do not have a good way of distinguishing GEMs with multiple cells. Currently we estimate that the doublet rate for VDJ libraries should be the same as GEX libraries.
It is expected that each cell-barcode typically contains one productive TRA and one productive TRB contig. Barcodes with multiple alpha/beta chains (in case of TCR) might throw some light onto the GEMs with more than 1 cells.
If multiple cells with the same clonotype have more than one alpha/beta chains, they are unlikely to be true doublets. The chance of multiple GEMs containing identical multiple chains is very low. If only one cell has multiple alpha or beta chains, it is hard to tell if it is a doublet versus a true biological phenomenon. The rate of true biological double-TRB or double-TRA cells is low. So the these barcodes could be a doublet, a barcode collision event, or a PCR chimera. But there is no way to tell with 100% certainty from this data.