WHY: doublet exclusion assumes height and area track together for single cells. The RATIO at which they track is instrument- and gain-dependent – on this cytometer the ridge sits near 0.56, not the 1.0 that a hardcoded gate would assume, and a band centred on 1.0 discards essentially every real cell. Centering on the measured median with a MAD-scaled width adapts to any instrument. MAD is used rather than SD because doublets are a heavy tail.
Details
ONE-SIDED. Two cells passing together give a pulse of about twice the area and much less than twice the height, so a doublet's height-to-area ratio is LOW (Wersto et al. 2001). Nothing about a doublet raises it. The gate therefore has a lower bound only; an upper bound removed single cells from the top of the ridge for no reason.
WHY NO SIDE-SCATTER STEP. SSC-H against SSC-A after the forward-scatter gate is common manual practice (Cossarizza et al. 2019), and was tried here with the same rule. Scored against manual gating on 21 whole-blood files it removed 3.8% of the low-scatter cells and 0.7% of the granulocytes: where little light is scattered the SSC ratio is noisy, and a cut below its median removes lymphocytes rather than doublets. Granulocyte shares rose by about a point against the manual counts, so it is not applied.