In-Place Transfers: Adjusting an Already-Seeded Plate

So far every example has filled empty target wells from scratch. Often the target plate may already have reagents in every well, and the next step is to add something to each well – adjusting pH, dosing a stimulant – without discarding what's already there. Pourfecto calls this an in-place transfer: the source and target are the same physical wells.

This example adds sodium hydroxide to every well of an already-filled plate to neutralize pH, keeping the existing water and buffer intact.

using Pourfecto, CHESSCore, Unitful

Reagents and labware

existing_plate represents the plate's current contents. target_plate represents what each well should contain afterward – the same water and buffer, plus the NaOH addition. Because Well.stock is a single mutable field, "current state" and "desired state" for the same physical plate have to be two separate Labware objects; Pourfecto recognizes them as the same plate by matching labware name and well name, not object identity, so both are built with the same name.

water = string_to_component("water", Liquid)
buffer = string_to_component("buffer", Liquid)
naoh = string_to_component("naoh", Liquid)

existing_plate = build_location(location_kinds[:WP96], "assay_plate")
for w in vec(children(existing_plate))
    w.stock = 150u"µL" * water + 50u"µL" * buffer
end

target_plate = build_location(location_kinds[:WP96], "assay_plate")
for w in vec(children(target_plate))
    w.stock = 150u"µL" * water + 50u"µL" * buffer + 5u"µL" * naoh
end

naoh_reservoir = build_location(location_kinds[:DeepReservoir])
children(naoh_reservoir)[1].stock = 100u"mL" * naoh

configs = ["single_channel", "eight_channel_horizontal", "plate_master"]

Each well already holds 200 µL against a 400 µL physical capacity (CHESSCore.wellcapacity), so there's comfortable headroom for the 5 µL addition – capacity isn't the constraint in this example, but it's the real bound Pourfecto checks for in-place wells, not the declared target total.

A first attempt: forgetting to opt in

existing_plate and target_plate share the name "assay_plate". Pourfecto treats a source/target name collision as a likely mistake by default – reusing a name changes the physical meaning of the transfer – so it's rejected unless you explicitly opt in:

try
    pourfecto([naoh_reservoir, existing_plate], [target_plate], configs)
catch e
    println(e)
end

Setting up the in-place transfer

allow_in_place=true opts in to matching same-named labware. Once matched, Pourfecto pins each existing well's content – it's forced to transfer at its full existing quantity into itself – so nothing is lost, and no other well can draw from it. NaOH gets priority 0 (must land exactly, no slack) since it's the only thing actually being dosed here.

priority = PriorityDict("naoh" => UInt64(0))

pc = pourfecto([naoh_reservoir, existing_plate], [target_plate], configs;
    allow_in_place=true, priority=priority)

transfers(pc) is the source-by-target volume matrix; rows follow the order sources were passed in: [naoh_reservoir, existing_plate]. For any well, the existing-plate row should show its full 200 µL carried forward, and the naoh_reservoir row should show exactly 5 µL.

Incomplete target declaration

Suppose the existing plate also has a trace of a preservative that was omitted in the target – it's not something the plan is trying to change, just an oversight:

preservative = string_to_component("preservative", Liquid)

existing_plate_2 = build_location(location_kinds[:WP96], "assay_plate_2")
for w in vec(children(existing_plate_2))
    w.stock = 150u"µL" * water + 50u"µL" * buffer + 2u"µL" * preservative
end

target_plate_2 = build_location(location_kinds[:WP96], "assay_plate_2")
for w in vec(children(target_plate_2))
    w.stock = 150u"µL" * water + 50u"µL" * buffer + 5u"µL" * naoh  # preservative omitted
end

This isn't a naming mistake – allow_in_place=true is set correctly – but it's still infeasible, for the same underlying reason explored in the priority example: a reagent that's never declared in any target defaults to priority 0 (blocked), which means its delivered amount must equal its target exactly – zero. The pin says the preservative's existing 2 µL must carry forward; priority 0 says it must be zero. Unlike the priority example, there's no alternate source to fall back on here, so this isn't a silent shortfall – it's a hard infeasibility, and Pourfecto's own diagnostics recognize this exact combination:

try
    pourfecto([naoh_reservoir, existing_plate_2], [target_plate_2], configs;
        allow_in_place=true, priority=priority)
catch e
    println(e)
end

The fix is simple: restate the full existing composition in the target, not just the part you're changing.

target_plate_2_fixed = build_location(location_kinds[:WP96], "assay_plate_2")
for w in vec(children(target_plate_2_fixed))
    w.stock = 150u"µL" * water + 50u"µL" * buffer + 2u"µL" * preservative + 5u"µL" * naoh
end

pc2 = pourfecto([naoh_reservoir, existing_plate_2], [target_plate_2_fixed], configs;
    allow_in_place=true, priority=priority)

The general lesson

An in-place well's entire existing composition needs to be restated in its target – not just the part you're adding or changing – or given explicit nonzero priority if you genuinely want it to be able to drift. Leaving something out doesn't mean "don't care" here any more than it did in the priority example; it means "blocked." The difference is that a fresh, empty target well often has another source it can fall back on when a reagent is blocked, so the result can be a silent inaccuracy. An in-place well is pinned to its own existing content with no alternative route, so the same mistake surfaces immediately as a hard infeasibility instead.

This page was generated using Literate.jl.