Improving flow cell utilization is the highest-return cost lever in a sequencing lab, and it is the one least likely to appear on anyone’s cost-reduction list, because the money it wastes is never itemized. A flow cell and its reagent kit cost essentially the same whether the run carries twelve samples or ninety-six, since the cost is in the run, not the samples on it. That single fact means the cost per sample is governed less by what you pay for reagents than by how full your flow cells go out the door, and it is why a lab that negotiates hard on reagent price while routinely running half-empty flow cells is economizing on the wrong number by an order of magnitude.
This article is about the three levers that control fill: batching, how you accumulate and group samples into runs; multiplexing, how many samples you combine on a flow cell; and scheduling, how you time runs across your samples and instruments. Together they set the effective cost per sample more powerfully than any purchasing decision, and this piece builds on the cost and turnaround models established elsewhere rather than repeating them.
Key Takeaways
|
The Economics of a Partly Filled Flow Cell
The cost structure of a sequencing run is what makes fill matter so much, and it is worth stating plainly because its consequence is so often missed. The dominant cost of a run, the flow cell and the sequencing reagent kit, is fixed the moment you start the run, independent of how many samples are loaded onto it. Add the per-sample library cost, which is roughly proportional to sample count, and the total cost per sample becomes the fixed run cost divided across however many samples share it, plus the per-sample library cost. Divide a fixed cost across fewer samples and the cost per sample rises, steeply, as the flow cell empties.
The table below shows the effect using representative figures for a flow cell whose run cost is on the order of several thousand dollars and whose capacity is a full plate of multiplexed samples. The exact numbers depend on your platform and pricing, but the shape of the curve is universal, and the shape is the point.
Samples on the Run | Flow Cell Fill | Sequencing Cost/Sample | Total Cost/Sample |
12 | ~12% | ~$500 | ~$580 |
24 | ~25% | ~$250 | ~$330 |
48 | ~50% | ~$125 | ~$205 |
72 | ~75% | ~$83 | ~$163 |
96 | ~100% | ~$62 | ~$142 |
Table 1. How cost per sample falls as flow cell fill rises, using a representative run cost and a per-sample library cost. Figures are illustrative planning values, not vendor pricing; the curve’s shape, not the exact numbers, is the point.
Fill Beats Reagent Negotiation, by an Order of Magnitude In this representative model, a half-full flow cell costs roughly 44% more per sample than a full one, $205 against $142. Scale that across a year: a lab running 2,000 samples at 50% fill instead of 100% spends on the order of $125,000 more, purely from empty lanes. A hard-fought 10% reduction in reagent price, by comparison, might save that same lab around $12,500 a year. The fill discipline is worth roughly ten times the reagent negotiation, and it requires no supplier concession at all, only the operational discipline to run full flow cells. This is why fill, not price, is where the cost-per-sample battle is actually won. The full flow-cell cost economics behind these figures are worked through in How Much Does NGS Cost? Budgeting Instruments, Reagents, and Sequencing. |
Setting Batch Size and Hold Windows
If full flow cells are so much cheaper per sample, the obvious question is why any lab runs partial ones, and the answer is turnaround. Filling a flow cell means waiting for enough samples to accumulate, and that wait is the cost of the saving. A batching policy is the deliberate resolution of this tension: it sets a target batch size, tied to flow cell capacity, and a maximum hold window, the longest a sample will wait for its batch to fill before the run goes ahead regardless. Together these two parameters decide where a lab sits on the tradeoff between cost per sample and turnaround.
The right settings depend on volume. At high volume, flow cells fill quickly and the hold window rarely binds, so a lab can run nearly full without waiting long; at low volume the wait to fill can grow long enough that a hold window has to cap it, accepting some partial runs to protect turnaround. That interaction between volume, fill, and waiting time, including the point at which the wait argues for outsourcing overflow, is modeled in the guidance on running a sequencing program at scale, linked below. The practical discipline is to set both parameters explicitly, as policy, rather than deciding run by run under pressure, so the cost-turnaround tradeoff is made deliberately rather than by whoever is waiting most impatiently for a given batch.
Lab Quality Management Certificate
The Lab Quality Management certificate is more than training—it’s a professional advantage.
Gain critical skills and IACET-approved CEUs that make a measurable difference.
Multiplexing Depth by Application
Multiplexing, combining many indexed samples on one flow cell, is what makes high fill possible, but the right number of samples to combine is set by a constraint that has nothing to do with the flow cell’s maximum capacity: the amount of data each sample needs. Every sample on a multiplexed run shares the flow cell’s total output, so the more samples you load, the less data each one receives. Multiplexing depth is therefore capped by the most data-hungry application in the pool, not by how many index combinations the flow cell can physically distinguish.
This is where multiplexing policy and the required-data calculation meet. A run of samples that each need deep coverage can only carry a few before each falls below its data requirement, while a run of samples needing shallow data can be multiplexed very deeply. Sizing the data each application actually needs, so that multiplexing depth is set correctly rather than by habit, is exactly what the Sequencing Coverage and Cost Calculator: How Much Data Do You Actually Need? is for. The policy point for a manager is simple: multiplex as deeply as the least-shallow sample in the pool allows, and group samples with similar data needs together so that a few data-hungry samples do not force a whole flow cell to run under-multiplexed.
Group by Data Need, Not Just by Arrival The subtle efficiency most labs miss is pooling samples with similar per-sample data requirements together. Mix a few samples needing deep coverage with many needing shallow data on the same flow cell, and the deep-coverage samples force a low multiplexing depth on the whole run, wasting capacity the shallow samples did not need. Sort samples into pools by their data requirement, and each flow cell can be multiplexed to the right depth for its contents. This is a scheduling and grouping decision, not a chemistry one, and it recovers fill that arrival-order pooling silently loses. |
Interested in lab leadership?
Turnaround Commitments That Survive Batching
A batching policy that optimizes fill will fail the first time it collides with a turnaround promise it cannot keep, so the turnaround commitments a lab makes have to be built on the batched reality of how it runs, not on an idealized fast case. Promising a turnaround that only holds if a flow cell happens to be full when a sample arrives is promising something the lab cannot reliably deliver, and the gap between the promise and the batched reality becomes a recurring source of missed commitments and strained relationships with submitting groups.
The honest approach is to define turnaround against the hold window: if the policy holds samples up to a defined maximum before running, the committed turnaround has to accommodate that maximum plus the run and analysis time, not assume it away. A lab that offers a realistic turnaround it can meet consistently serves its users better than one that offers an optimistic turnaround it meets only when batching happens to cooperate. Where a subset of work genuinely needs faster turnaround than the batch policy allows, that is an argument for a defined priority path, or for outsourcing that overflow, rather than for quietly breaking the batching discipline that keeps the whole operation economical.
Scheduling Across Multiple Instruments
A lab with more than one instrument gains a scheduling lever that a single-instrument lab does not have, and using it well can lift fill across the whole operation. With multiple instruments, runs can be arranged so that each goes out as full as possible rather than spreading the same samples thinly across several partial runs, and instruments can be assigned to different run types or turnaround tiers, one kept available for urgent work while another runs large batched production loads at maximum fill. The goal is to keep every started run as full as it can be, rather than keeping every instrument merely busy.
This is where a published run schedule earns its value, because it lets submitting groups plan backward from known run slots, which naturally improves batching without per-run negotiation and lets the lab consolidate samples onto fewer, fuller runs. The broader scheduling and throughput discipline this fits within, including how to diagnose whether fill is even the constraint before optimizing it, is covered in Running NGS at Scale: Throughput, Scheduling, and Automation. Across batching, multiplexing, and scheduling alike, the through-line is the same: the flow cell costs what it costs whether it runs full or empty, so every lever that fills it is a lever on cost per sample, and together they matter more than any price you negotiate. How this fits the full program economics is in Next-Generation Sequencing in the Lab: A Manager’s Guide to Building, Budgeting, and Scaling NGS Capacity.
This article was produced under Lab Manager's AI Editorial Guidelines.
