How the gauge works

Every configuration carries three numbers, and they put the marker count into one of four bands.

On a conventional instrument the detector count is the answer. One marker needs one detector, so a 16-detector configuration tops out at 16 colours and usually at 14 or 15 once a viability dye and some headroom for spreading are taken out. The ceiling here is hardware and the tool treats it as hard.

On a spectral instrument the detector count tells you almost nothing. An Aurora 5L reads 64 detectors and nobody runs 64 markers. What limits a spectral panel is how far apart the full spectral signatures are, so the ceiling comes from what people have actually published on that configuration rather than from the hardware. That number is a judgement. It is the part of this tool most worth arguing with.

On a conventional instrument the detector count splits across the lasers. A FACSCanto II reads eight colours, but only two of them through the violet laser and four through the blue. Four violet dyes cannot go through two violet detectors however far apart their emissions sit, so the planner treats a full laser as a hard stop rather than a penalty.

A spectral instrument has no such split. Every detector sees every laser, so only the total matters there and this rule does not apply.

Selecting a population selects every gate above it, because you cannot report a subset without resolving its parents. Central memory CD4 T cells need CCR7 and CD45RA, and they also need CD4, CD8, CD3 and a viability dye, because that is the path the gate takes to get there. The marker table shows the union across everything you picked, which is why adding a second subset of a lineage you already have often costs nothing.

Scatter and time are gates too, and they never consume a detector, so they are excluded from the count. Viability is not excluded. It takes a channel like anything else, and panels that forget it come up one short.

The suggestion runs greedily over the markers in order of how constrained they are, dim antigens first. Bright antigens can go almost anywhere, so letting them take the bright dyes early strands the markers that had no alternative. Assigning in that order is what keeps CD25 and CD127 off the leftovers.

Three penalties sit on top of the brightness match. A tandem dye on an intracellular or nuclear target is marked down, because fix and perm degrade tandems. A dye whose emission maximum sits within 18 nm of something already placed on the same laser is marked down, which is a crude proxy for spreading. And polymer dyes are marked down once a panel already leans on several of them.

The last step matches the panel to conjugates somebody actually sells. The catalogue records one row per marker, species, vendor and clone, listing the fluorochromes that clone is sold conjugated to. Catalogue numbers are shown where they were verified, but they change often enough that the colour list is the part worth trusting.

Because a clone is only sold in its own set of colours, choosing one can invalidate a dye the planner already picked. Changing a clone therefore re-plans the panel rather than leaving a pairing nobody sells.

When an assignment strands a marker with nothing orderable, the planner reshuffles and tries again. It runs four different orderings and keeps whichever produces the fewest unsourceable markers. The one that usually wins puts the most constrained marker first: a marker sold in three colours has to choose before one sold in thirty.

The mark beside each reagent says how much anybody checked, in three levels. A catalogue number means that exact conjugate is confirmed. A bare vendor name means somebody confirmed the vendor sells that clone, without looking up this particular colour. A question mark means nobody cited it at all. About three in a hundred pairings carry a number, so the middle level is where most of the catalogue sits.

Every row records whether anyone could cite it. A vendor shown plainly traces to a product title or a catalogue number. A vendor followed by a question mark is a plausible format for that clone that nobody could point to a source for, which is a good guess and not a fact. The distinction is kept because a conjugate that does not exist is worse than a gap: it produces a panel that looks finished and cannot be bought.

A marker missing from the catalogue means unknown, not unavailable. The catalogue is curated rather than exhaustive, so a marker it has never heard of is left unconstrained and flagged. Treating a gap in the data as proof that nothing exists would quietly throw away panels that are perfectly buyable.

It does not calculate spillover or a spreading matrix. The 18 nm emission check is a warning, not a measurement, and it will miss pairs that spread badly despite being far apart. Run the real similarity index in your vendor's tool before you order anything.

It does not price anything, check stock, or know about your institutional supplier agreement. It also does not cover every marker: where the catalogue is silent the panel is marked unchecked rather than approved, and you should confirm those conjugates yourself.

It does not know your sample. Antigen density is stored as a single value per marker, and in real life it moves with activation state, tissue and fixation. Where a marker's density decides the panel, check it on your own cells.

The population trees are transcribed from published consensus panels and each one carries its source and a verified flag. A tree marked unverified was assembled from general knowledge and needs checking before anyone leans on it.