Hi Daniel,
For compressed barcoding, it matters that the fraction of positive samples is low, because otherwise all sub-barcodes will become positive, precluding deconvolution. Our simulations of false-positive and false-negative rates indicate that if 1.3% of samples were positive, compressed barcoding would be feasible with FPR/FNR <0.2% with the indicated parameters (10k total barcodes, 5 barcodes per sample, 3/5 detected, 10 pools per run).
Best,
Jonathan