Flow Cytometry interview questions

What a strong answer covers

• Forward scatter is light scattered at a small angle and correlates broadly with cell size; side scatter is light collected at right angles and reflects internal complexity or granularity.
• Together they give a label-free first look at the sample: in peripheral blood, lymphocytes, monocytes and granulocytes occupy recognisably different regions, and debris sits low on forward scatter.
• The first practical use is excluding debris and cell fragments so they do not dilute later frequencies, remembering that a low forward scatter event can be a small real cell rather than junk.
• Area, height and width of the pulse carry different information, which is what makes doublet discrimination possible from the same signals.
• A strong answer notes these are relative measures on arbitrary axes, not calibrated size readings, so comparisons across instruments and settings need care.

Where people lose the point

• ×Calling forward scatter a direct measurement of cell diameter.
• ×Gating out low scatter events without checking whether a small population of interest lives there.
• ×Assuming the scatter profile of one tissue applies to another, for example expecting blood-like populations in a dissociated solid tissue.

What a strong answer covers

• A doublet is two cells passing the laser as one event, so it is recorded with roughly double the signal but a longer pulse. Comparing pulse area against pulse height (or width) separates them, since area grows with the longer pulse while height does not scale the same way.
• Doublets matter because they manufacture false double positives: a marker-A cell stuck to a marker-B cell reads as a single A and B positive event, which is exactly the population people over-interpret.
• Doublet exclusion is standard practice in sorting too, because a sorted doublet delivers the wrong cell along with the right one and quietly reduces purity.
• Good practice is to exclude on more than one scatter parameter and to check the excluded fraction: an unusually high doublet rate is telling you about sample preparation, not just about the gate.
• A strong answer connects the fix upstream: proper dissociation, filtering the sample before acquisition, appropriate concentration and, where relevant, a DNase or EDTA step for clumpy samples.

Where people lose the point

• ×Skipping doublet exclusion and then reporting a rare double-positive population.
• ×Drawing the exclusion gate so tightly that real cells with variable pulse shapes are discarded.
• ×Treating a high doublet rate as an analysis nuisance rather than a signal that the sample needs filtering or diluting.

What a strong answer covers

• Fluorophores emit over broad spectra, so light from one dye is detected in more than one detector. Compensation is the arithmetic that removes the average amount of that spillover so a signal in a detector reflects its own fluorophore.
• A valid single-stain control uses the exact same fluorophore (and, for tandem dyes, the same conjugate lot where possible), is at least as bright as the sample, and has a negative population carrying the same autofluorescence as the positive population it is compared against.
• That autofluorescence requirement is why beads and cells must not be mixed within one control: the negative from a bead control cannot serve as the negative for a cell-stained positive.
• Compensation values should be calculated by the software from the controls, not dialled in by eye until the plot looks right, and they should be checked with the medians of the positive and negative populations rather than by appearance.
• What compensation does not fix is spillover spread: the extra measurement error carried in with the spillover, which widens populations even when the correction is exactly right.
• A strong answer distinguishes compensation on a conventional cytometer from spectral unmixing on a spectral instrument, where the full emission signature is fitted rather than a pairwise matrix applied, while noting that unmixing still needs good single-colour references and still cannot remove spread.

Where people lose the point

• ×Adjusting compensation manually until populations look square, which typically over-compensates and pushes real dim positives into the negative.
• ×Using a control whose fluorophore is not identical, most often substituting one tandem conjugate for another.
• ×Blaming compensation for a fanned-out population that is actually spillover spread.

What a strong answer covers

• An FMO tube contains every fluorophore in the panel except the one you are gating on, so it shows exactly where the signal in that detector begins once all the spillover and spread from the rest of the panel is present. That makes it the right control for placing a gate in a multicolour panel.
• An isotype control substitutes an irrelevant antibody of the same isotype, host species and ideally the same conjugate and fluorophore to protein ratio, and it addresses non-specific binding rather than spillover.
• FMO has become the default because in modern panels the boundary between negative and dim positive is usually set by spread from other colours, which an isotype tells you nothing about.
• Isotypes remain useful when non-specific binding is genuinely suspected, for example on cells with high Fc receptor expression, but a poorly matched isotype gives a false sense of rigour and can be worse than no control at all.
• Neither replaces Fc receptor blocking, and neither replaces a real biological negative population, which is the strongest control available when the biology provides one.
• A strong answer says it plainly: FMO for gate placement, isotype for a specific non-specific binding question, and internal negative populations wherever the sample supplies one.

Where people lose the point

• ×Describing FMO and isotype as interchangeable, or gating positivity off an isotype in a large panel.
• ×Using an isotype that differs in fluorophore, clone format or conjugation ratio and treating it as matched.
• ×Forgetting that an FMO must be run on the same sample type and stained in the same conditions as the test tube.

What a strong answer covers

• Order matters and should be justified: time or acquisition stability first if the instrument had any fluidics disturbance, then scatter to remove debris, then doublet exclusion, then a viability gate, then lineage markers, then the phenotype of interest.
• Dead cells come out before phenotyping because they bind antibody non-specifically and are a common source of apparent positivity in every channel.
• Each boundary should be justified by something other than appearance: an FMO, an internal negative population, or a biologically defined reference such as a known negative lineage.
• Report the parent population for every frequency you quote. A number is meaningless without saying what it is a percentage of, and comparing frequencies with different parents is a frequent source of disagreement between labs.
• Back-gating is the defensibility check: take the final population and display it on the earlier plots to confirm you did not amputate part of it upstream.
• A strong answer mentions consistency across samples in a study: the same strategy applied to every file, with any per-sample gate adjustment documented and justified rather than made silently.

Where people lose the point

• ×Drawing gates by eye to make the result match the hypothesis, and adjusting them per sample without documenting it.
• ×Phenotyping before excluding dead cells and doublets.
• ×Quoting percentages without stating the parent gate.

What a strong answer covers

• Start from the instrument, not the antibody catalogue: the lasers, filters and detector configuration decide which fluorophores are usable at all.
• Match brightness to antigen density in the opposite direction from intuition: the dimmest, most poorly expressed antigens get the brightest fluorophores, while highly expressed lineage markers can afford dim ones.
• Use the spillover spreading matrix to place markers, keeping a dim marker out of detectors that receive heavy spread, and try to avoid putting two markers that are co-expressed on the same cell into a badly spreading pair.
• Treat tandem dyes with caution: they are susceptible to degradation from light, fixation and temperature, and they vary by lot, so their compensation controls should come from the same lot and be freshly prepared.
• Plan the controls at the same time as the panel: single stains for every fluorophore, FMOs for the markers whose gates are not obvious, and a viability dye compatible with any fixation or permeabilisation you intend.
• A strong answer describes the validation stage: titrate every antibody rather than following the datasheet volume, test the panel on a known sample, and compare each marker's resolution in the full panel against how it looked alone.

Where people lose the point

• ×Putting the brightest fluorophore on the most abundant marker, which wastes it and creates spread everywhere.
• ×Skipping titration and using the manufacturer's suggested volume, which usually costs resolution and money at once.
• ×Adding a colour without checking whether the instrument configuration can actually distinguish it from what is already in the panel.

What a strong answer covers

• The core distinction is whether the dye survives your workflow. DNA-binding dyes such as propidium iodide or 7-AAD stain cells with compromised membranes and are appropriate for live, unfixed acquisition, but the signal is not reliable once cells are fixed and permeabilised.
• Amine-reactive fixable dyes covalently label free amines. Dead cells label far more intensely because the dye reaches the whole protein content, and because the label is covalent it survives fixation and permeabilisation, which is what makes it the choice for intracellular staining panels.
• The choice also has to fit the panel spectrally: a viability dye occupies a detector, so it competes with markers and contributes spillover like any other colour.
• Excluding dead cells matters because they bind antibodies non-specifically and are autofluorescent, which produces apparent positivity spread across channels and inflates rare-population frequencies.
• A strong answer mentions timing and buffer requirements, particularly that amine-reactive dyes should be applied in protein-free buffer, since protein in the buffer competes for the dye.

Where people lose the point

• ×Using a DNA-binding dye in a panel that will be fixed and permeabilised.
• ×Staining with an amine-reactive dye in a buffer containing serum or BSA, which quenches the reaction.
• ×Reporting a rare population without a viability gate and attributing the signal to biology.

What a strong answer covers

• Localise the problem before changing anything: is it this marker only, this tube only, this sample type, or every file from today. That single question separates a staining issue from an instrument issue.
• Check the instrument first if it is global: QC beads against the established baseline, laser delay and detector performance, and whether voltages or the configuration were changed since the last good run.
• If it is one marker, work through the reagent: antibody age and storage, tandem degradation, whether the conjugate lot changed, and whether the titre is still right for the current cell number.
• Then the sample: viability, over-digestion during dissociation, storage time before staining, and whether the antigen is sensitive to the fixation or permeabilisation used.
• Check the biology last but do not skip it: down-regulation or internalisation of the target after stimulation is a real cause, and the assay may be reporting the truth.
• A strong answer keeps a record of the comparison against a known good file, and mentions that comparing the stain index for that marker across runs turns a subjective impression of dullness into a number.

Where people lose the point

• ×Increasing detector voltage to make the population look separated, which moves the picture without improving resolution.
• ×Changing more than one thing at once so the cause is never identified.
• ×Assuming the cytometer is at fault without running instrument QC to check.

What a strong answer covers

• Sorting adds a decision under time pressure: the instrument must classify and deflect each drop, so gate placement now determines what physically ends up in the tube rather than what appears in a plot.
• Purity and yield trade against each other through the sort precision mode. A stricter mode rejects any drop with a conflicting neighbour, raising purity and discarding more target cells; a permissive mode keeps more cells and admits more contaminants.
• Which way to lean depends on the downstream use: single-cell sequencing or culture initiation usually demands purity, while a bulk assay on an abundant population may accept contaminants in return for enough material.
• Nozzle size, sheath pressure and sample concentration have to suit the cells: large or fragile cells need a bigger nozzle and lower pressure, and running the sample too concentrated increases conflicts and drops both purity and yield.
• Sterility, temperature and collection medium matter for anything going back into culture, and post-sort viability is part of the result, not an afterthought.
• A strong answer always ends with a post-sort purity check on an aliquot, and a record of what the sort actually achieved rather than what was requested.

Where people lose the point

• ×Promising both maximum purity and maximum recovery without acknowledging the trade.
• ×Running the sample too concentrated to save time, then reporting disappointing purity.
• ×Not checking post-sort purity and viability, so a failed sort is discovered downstream.

What a strong answer covers

• Acquisition volume becomes the limiting factor: you need enough total events for the target count to be statistically meaningful, so the plan should start from how many target events are required and work backwards.
• The tolerance for background collapses. At very low frequencies, non-specific binding, dead cells, doublets and spillover contribute events that are indistinguishable from a small real population, so exclusion gating has to be stricter, not looser.
• Adding a dump channel that removes irrelevant lineages, and using more than one positive marker for the target, gives the population a signature rather than a single-parameter claim.
• Controls must include something that establishes the background rate: a sample known not to contain the population, processed identically, tells you the floor below which a result is not interpretable.
• Carryover between tubes is a real contributor at these frequencies, so wash steps between samples and running a blank after a strongly positive sample are worth the time.
• A strong answer states a limit of detection for the assay rather than reporting whatever percentage the software prints.

Where people lose the point

• ×Reporting a frequency computed from a handful of events without a confidence interval or a stated detection limit.
• ×Using the same gating stringency as a routine phenotyping run.
• ×Ignoring carryover after acquiring a strongly positive sample immediately before the rare-event one.