Pooling donors is usually presented as a way to reduce variability, which gets the statistics exactly backwards. Pooling does not reduce variability. It hides it, converts it into a fixed and unmeasurable property of your reagent, and removes your ability to ask whether the effect you are seeing depends on who the cells came from.
What pooling actually does
Combining cells from several donors produces a preparation whose behavior is some composite of its inputs. That composite is stable across the batch, which is why it feels like consistency. What has happened is that between-donor variation moved out of your dataset and into your reagent, where it is no longer visible and no longer measurable.
Averaging is genuinely useful when donor identity is noise. It is destructive when donor identity is signal, and in immunology it is signal far more often than protocols assume.
Where pooling is defensible
| Situation | Pooled | Single donor |
|---|---|---|
| Early screening for a gross effect | Reasonable | Unnecessary at this stage |
| Method development | Reasonable, a consistent substrate is useful | Not required |
| Anything donor-dependent | Actively misleading | Required |
| HLA or KIR restricted work | Meaningless, the genotypes are mixed | Required |
| Allogeneic product development | Does not represent a real product | Required |
| Understanding population variance | Removes the thing you are studying | Required, across several donors separately |
The distinction is whether you need an average or a distribution. Screening wants an average. Development wants a distribution, because the distribution is what tells you whether a result will hold in the next donor.
The mixed-lymphocyte problem
There is a specific reason pooling immune cells is different from pooling most reagents. Lymphocytes from different donors recognize each other as foreign. Put them together and they begin responding to one another, which means a pooled preparation of immune cells is not a passive mixture, it is an ongoing reaction.
For phenotyping over short windows the effect may be tolerable. For any culture-based or functional work it introduces activation that has nothing to do with your experimental question and cannot be subtracted afterwards.
The false economy
Pooled material often looks cheaper per experiment, and for a screening campaign it may genuinely be. The economics reverse the moment a result needs to be defended, because the first question asked of an interesting effect is whether it generalises, and a pooled preparation cannot answer it.
The middle path most programs should be on
Use a defined panel of individually characterized single donors, run separately, with results reported per donor rather than averaged into a single figure. This gives the robustness pooling is reaching for while keeping the variance visible, and it costs the discipline of tracking donor identity through the analysis rather than dropping it at the point of purchase.
It also produces something a pooled design cannot: a documented range. Knowing that an effect spans a particular spread across five characterized donors is a far stronger claim than a single number from a mixture of unknown composition.
Sourcing for it
OrganaBio documents single-donor material across the catalogue, including fresh and cryopreserved leukopaks at a minimum of 10 billion cells per single-donor unit, cryopreserved PBMCs, and isolated T and NK populations, with characterization applied at donor program level including high-resolution NGS HLA typing across six genes and KIR genotyping. Eligible donors can be scheduled for repeat collection, which is what turns a set of single donors into a panel you can return to. Building one is covered in HLA-matched material for allogeneic programs.
Frequently asked questions
Does pooling donors reduce variability?
No, it hides it. Between-donor variation moves out of your dataset and into your reagent, where it is fixed, invisible and unmeasurable. That feels like consistency but removes your ability to ask whether an effect depends on donor.
When is pooled cell material appropriate?
Early screening for a gross effect and method development, where a consistent substrate is useful and donor identity is genuinely noise. It is inappropriate for anything donor-dependent, HLA or KIR restricted work, allogeneic product development, or any study of population variance.
Why is pooling immune cells specifically problematic?
Lymphocytes from different donors recognize each other as foreign and begin responding to one another. A pooled immune cell preparation is an ongoing reaction rather than a passive mixture, which introduces activation unrelated to your experimental question.
Are three runs on pooled material the same as three donors?
No. Three runs on a pooled preparation are technical replicates of one composite. Three single donors are biological replicates, and only those tell you whether a finding generalises beyond a single mixture.
Is pooled material cheaper?
Often per experiment, and for a screening campaign that may be real. The economics reverse when a result needs defending, because the first question asked of an interesting effect is whether it generalises and a pooled preparation cannot answer it.
What is the alternative to pooling?
A defined panel of individually characterized single donors, run separately, with results reported per donor rather than averaged. It gives the robustness pooling reaches for while keeping variance visible, and produces a documented range rather than a single number.
What makes a set of single donors into a usable panel?
Characterization applied before selection, so donors can be chosen on genotype rather than filtered afterwards, and the ability to return to those same donors over time through a repeat-collection program.
Talk to OrganaBio
Working through this on a live program?
The scientific team works through sourcing and specification questions with cell therapy and research groups directly, including donor characterization, format selection and documentation scope.
