Even if technological improvements have cut down the frequency, I imagine there could be user error or other circumstances that would increase the number of barcode multiplets in a sample. I figure they would be identified as "non-empty" barcodes with high(er) probabilities at lower UMIs (like in the plot below) and should not affect the output. But I was wondering if anyone has already tested this with simulated data, artificial multiplets or some other way.
Has anyone explored barcode multiplets data using this resource? Lareau et al. provided some compelling results regarding this (https://www.nature.com/articles/s41467-020-14667-5).
Even if technological improvements have cut down the frequency, I imagine there could be user error or other circumstances that would increase the number of barcode multiplets in a sample. I figure they would be identified as "non-empty" barcodes with high(er) probabilities at lower UMIs (like in the plot below) and should not affect the output. But I was wondering if anyone has already tested this with simulated data, artificial multiplets or some other way.