Recurring contamination has a familiar rhythm. Something appears where it should not, everything is discarded, fresh reagents are opened, the bench is cleaned thoroughly, and the problem goes away for a fortnight before returning exactly as before. Each round consumes a week and teaches nothing, because replacing everything at once cannot identify which thing was responsible.
The reason this cycle persists is that the standard response is elimination by exhaustion rather than by evidence. It works occasionally and by accident, and when it fails the laboratory has no more information than it started with.
There is a better method, and it comes from epidemiology. Outbreak investigators face the same structural problem: an agent appearing in some places and not others, many possible routes, and no ability to see the transmission directly. Their approach is to characterise the agent precisely, enumerate every plausible exposure, then design comparisons where a single factor differs. That method transfers to a bench problem almost unchanged.
Key takeaways
- Characterise the contaminant before changing anything, because the signature constrains the source list.
- Map every shared reagent, surface and instrument the affected and unaffected work have in common.
- Split reagent lots and test halves in parallel to convert guesswork into a controlled comparison.
- Sequencing identifies microbial contaminants and frequently names the reservoir directly.
- Replacing everything at once cures the symptom and destroys the evidence.
Defining the Contamination Signature
The first discipline is to stop changing things. Every replacement made before the contaminant is characterised destroys evidence, and a source that has been unknowingly discarded leaves the laboratory unable to explain a recurrence months later.
Characterisation starts with what the contaminant is. In molecular work, an unexpected amplification product has a size, a melting profile and a sequence. In cell culture, contamination has a morphology and a growth rate. In analytical chemistry, an unexpected peak has a retention time and, with mass spectrometry, a fragmentation pattern that often identifies the compound outright. Each is a fingerprint, and a fingerprint constrains the list of plausible sources far more than a vague sense that something is wrong.
Next comes the pattern of occurrence, and this is where most of the information sits. Four questions matter.
Which samples are affected, and which are not? A contaminant present in every reaction including the no-template control points to a reagent or the environment. One appearing only in samples from a particular source points upstream to collection or preparation. One confined to specific plate positions points to physical transfer during handling.
When did it start, and what changed then? An abrupt start usually coincides with a discrete event: a new reagent lot, a new consumable supplier, an instrument service, building work, a change in room air handling. A gradual appearance suggests something accumulating, such as an aerosol reservoir or a developing biofilm.
Where does it appear? Mapping affected work against bench, instrument, hood and room often reveals a spatial pattern pointing straight at a physical reservoir.
Who was involved? This must be handled as a technical variable rather than an accusation. Different people use different consumables, pipettes and step sequences, and a contaminant confined to one person’s work is usually about a piece of equipment or a step in their routine rather than carelessness.
Finally, quantify. A steady low level suggests a reagent present in every reaction; a highly variable one suggests intermittent transfer, which is more characteristic of aerosol or contact contamination. Writing all of this down before touching anything is the highest-value step in the process.
Mapping Every Shared Reagent and Surface
With a signature and a pattern established, the next step is to enumerate everything the affected work touched, and, equally important, everything the unaffected work touched. The comparison between those two lists is what narrows the field.
The map must be exhaustive and written down rather than held in memory. Cover water sources, buffers and stock solutions with their preparation dates, enzymes and master mixes, primers and probes, consumables down to tip and tube lots, gloves, pipettes and their ejectors, centrifuges, vortex mixers, thermal cyclers, hoods, incubators, waste containers, and the surfaces on which each step occurs.
Two categories are consistently under-mapped. The first is shared equipment nobody thinks of as part of the workflow: a centrifuge in another room, a shared plate reader, an ice machine. The second is reagents made in-house from other reagents, since a contaminated stock propagates silently into everything derived from it, and a working solution that looks clean by its preparation date may inherit contamination from a stock made months earlier.
Then sort each item into three groups. Common to affected work and absent from unaffected work: the strongest candidates. Common to both: weaker, though not excluded, since exposure alone does not guarantee an effect. Present only in unaffected work: effectively excluded. That sorting usually shrinks forty items to a handful, on paper in an afternoon rather than at the bench over weeks.
| Occurrence pattern | Most likely source category | First test to run |
|---|---|---|
| Every reaction including blanks | A shared reagent or the water used to make it | Test each reagent individually against clean stock |
| Only some batches, unpredictably | Aerosol or contact transfer during handling | Surface and air sampling, plus workflow observation |
| Only one operator’s work | An item or step unique to that routine | Swap equipment and workspace between operators |
| Only after a specific instrument step | Reservoir inside that instrument | Sample the instrument interior and run a blank through it |
| Started abruptly on a known date | A change made at that time | Revert the single change and observe |
The table is a starting point rather than a rule. Its value is in forcing an explicit hypothesis before any test is run, which is what distinguishes investigation from random replacement.
Splitting Reagent Lots as a Test
The central experimental technique is straightforward and underused: divide a reagent set, change exactly one thing between the halves, and run both in parallel.
Suppose eight candidate reagents remain. Testing them one at a time takes eight rounds. A better design tests groups. Run the full set alongside a set in which half the components have been replaced with freshly opened stock. If contamination follows the original half, the source is in that group and the group can be split again. Each round halves the candidate list, so eight candidates resolve in three rounds rather than eight.
Several details separate a clean answer from an ambiguous one.
Replace with genuinely independent material. A fresh tube from the same lot is not an independent replacement, because lot-level contamination is common; use a different lot and ideally a different supplier.
Change one variable per comparison. Replacing several suspect items at once defeats the purpose, since a negative result then implicates the whole group with no resolution.
Include the controls that make the result interpretable. A no-template control detects reagent contamination. An extraction blank carried through the full preparation detects contamination introduced during processing. A blank prepared in a different room separates reagent contamination from environmental contamination.
Repeat before concluding, because intermittent contamination produces clean rounds by chance. Two or three consecutive clean rounds, with the contaminated condition still producing the contaminant in parallel, is a real result.
Keep the contaminated material, labelled and sealed, until the investigation closes. It is the positive control that proves the fix worked, and discarding it is the commonest way an investigation ends inconclusively.
Sequencing Contaminants for Identification
When the contaminant is biological, sequencing frequently identifies the source directly, and it is often cheaper than another fortnight of guessing.
For an unexpected amplification product, sequencing it and comparing against reference databases answers the first question: is it a genuine organism, a primer artefact, or carryover of a previously amplified product? Each answer points differently. A primer dimer means the assay design or conditions need attention rather than the laboratory being dirty. Carryover, recognisable because it matches a target the laboratory has amplified before, points at amplicon contamination, the most stubborn category in molecular work because amplified product exists at enormous copy number and travels readily on aerosols and gloves.
Where a genuine organism is identified, the identity is informative. Certain genera recur as reagent-associated contaminants, particularly organisms adapted to nutrient-poor water, which is why purified water systems are a standing suspect in low-biomass work. Others are skin and respiratory flora, pointing to human contact and inadequate barrier practice. Others are environmental moulds and spore-formers, pointing to air, dust or building work.
For cell culture, short tandem repeat profiling identifies a cell line and detects cross-contamination by another, a widespread and under-recognised problem. Mycoplasma testing covers an organism invisible under a light microscope that passes conventional filters and alters cell behaviour substantially before anyone suspects it.
One caution matters greatly. In low-biomass sequencing work, contaminant sequences from reagents can dominate the result entirely, and the reagent contamination profile can be mistaken for a genuine finding. Processing blanks through the full workflow and sequencing them alongside samples is not optional in this setting. It is the only way to know which sequences came from the sample and which came from the kit.
Air Handling and Equipment as Reservoirs
When reagents have been excluded, the source is usually a physical reservoir that reseeds the workspace after each cleaning, and these are the sources that produce the fortnightly recurrence pattern.
Air movement is the most commonly overlooked. Rooms have pressure relationships and airflow patterns, and a room at positive pressure pushes air outward while one at negative pressure draws it in. Contamination travels along those paths, particularly where a pre-amplification area sits downstream of a post-amplification one. The correct arrangement is unidirectional workflow, clean areas upstream and amplified material downstream, with staff moving one way only.
Filters and cabinets deserve inspection: a cabinet with a loaded filter, a failed alarm or a lapsed certification no longer provides the protection it appears to. Building work nearby is a classic trigger, because disturbing ceiling voids and wall cavities releases accumulated spores into the air handling system.
Instruments harbour contamination internally. Thermal cycler lids and blocks accumulate residue from leaking tubes. Centrifuge rotors collect material from cracked tubes and are rarely opened. Liquid handling systems have wash lines and tubing where biofilm develops. Water baths are a known reservoir, and so are ice machines, since ice is rarely treated as a reagent although it contacts samples directly.
Pipettes are the single most common piece of equipment implicated in molecular contamination, because aerosol drawn into the barrel deposits inside the shaft and is expelled into subsequent samples. Filter tips address this, and only if used consistently. Pipettes should be dismantled and decontaminated on a schedule rather than only when a problem appears.
Sampling the environment converts suspicion into evidence. Contact plates and surface swabs, settle plates left open to sample air, and wipes from inside instruments and from door handles, keyboards, fridge handles and chair backs all build a map. Sampling before and after cleaning beats either alone, because a site clean after cleaning and contaminated again three days later is a reservoir rather than a residue.
Personnel Movement and Cross-Bench Transfer
People are effective transfer vectors, and the movements that matter are usually so routine that nobody notices them.
Gloves are the primary route. A gloved hand touches a contaminated surface, then a clean one, then a tube, and the chain is complete. The failure is not the absence of gloves but the failure to change them at transitions between areas or tasks, and watching an actual workflow rather than reading the procedure usually reveals several such transitions.
Objects travel with people. Notebooks, pens, phones, sample racks, ice buckets, tube boxes and laboratory coats all move between areas, and any of them can carry contamination across a boundary the workflow was designed to protect. Dedicating equipment to each area and never allowing items upstream is the standard control, and it fails most often for small shared objects rather than large obvious ones.
Direction matters as much as movement. Where clean and dirty areas are designated, the rule is that staff move one way during a session and do not return upstream without a break, a change of coat and a change of gloves. In practice people go back for a forgotten item, and that single return can undo the whole design. Shared equipment in a common area creates a hub where everyone’s contamination meets, connecting a clean and a post-amplification workflow regardless of how carefully the benches are separated.
Observation is the only reliable way to find these routes, and it needs careful framing. The aim is to map how work actually happens, which always differs from how it is documented, and the differences are usually sensible adaptations rather than laziness. An investigation that feels like a search for blame produces a sanitised performance, and the real workflow is precisely what needs to be seen.
Documenting the Fix So It Holds
An investigation that ends when the contamination stops has done half the work, because the same problem will return once the people involved have moved on.
Record the evidence, not just the conclusion. The contaminant’s signature, the occurrence pattern, the exposure map, the comparisons run and their outcomes, and the retained material all belong in a written record. A future recurrence then becomes a matter of comparing signatures rather than starting from nothing.
Distinguish the immediate fix from the systemic one. Discarding a contaminated reagent lot is the fix; changing how that reagent is received, aliquotted, stored and dated so a contaminated lot cannot propagate is the systemic action.
Verify rather than assume. Keep running the controls that detected the problem, at the same frequency, for long enough to be confident. Contamination recurring on a two-week cycle is not disproven by one clean week.
Build detection into the routine. Most laboratories discover contamination late because nothing was watching for it. Blanks carried through the full workflow, run on every batch and plotted over time, turn contamination into something caught within a run rather than noticed after a month of unexplained results.
Convert the finding into a control that does not depend on memory. If the source was pipette barrels, a scheduled decontamination with an owner and a date is the control. If it was directional workflow, a physical arrangement making the wrong direction inconvenient beats a written instruction. Controls relying on people remembering are the ones that quietly lapse.
Finally, share the finding. Contamination sources are frequently shared across groups using common facilities, and a water system, an ice machine or a shared instrument implicated in one laboratory’s problem is likely feeding another’s. A short written account circulated to neighbours costs an hour and occasionally saves someone else a month.
Frequently asked questions
Should everything be replaced when contamination appears?
Not before the contaminant has been characterised and the exposure map written down. Replacing everything at once frequently works, but it identifies nothing, so the same problem returns and the same expensive response repeats. The exception is where the affected work is genuinely urgent and cannot wait, in which case replace what is needed to continue, but retain and label everything removed so that the investigation can still be run afterwards on the original material.
How can a reagent be contaminated straight from the supplier?
Reagents are manufactured and packaged in facilities that are clean but not sterile, and several classes are known to carry low-level microbial genetic material from raw materials or from the water used in production. This matters enormously in low-biomass applications, where the sample contains little target and reagent-derived sequences can outnumber genuine ones, and hardly at all in high-biomass work, where the same trace contamination is swamped. The practical response is to process blanks through the entire workflow and record which lots were used for which experiments.
What distinguishes amplicon contamination from other kinds?
Scale and persistence. A completed amplification reaction holds an extraordinary number of copies of a short, robust molecule, so opening a tube releases far more target material than any biological sample contains. It settles on surfaces, travels on gloves and equipment, and resists ordinary cleaning because the molecule is stable. This is why the standard control is architectural rather than procedural: physically separate areas for preparation and post-amplification handling, dedicated equipment in each, unidirectional movement, and where the assay permits, enzymatic systems that render carried-over product unamplifiable.
Is bleach the right thing to clean with?
For nucleic acid contamination, a fresh dilute hypochlorite solution is effective because it degrades the molecule rather than merely dislodging it, and alcohol alone does not do this. Two caveats apply. The solution must be freshly prepared, since hypochlorite loses activity in dilution over days. And residue must be removed afterwards with water or alcohol, both because it corrodes equipment and because carryover can inhibit downstream reactions. For microbial contamination the appropriate agent depends on the organism, and spore-formers in particular resist many common disinfectants.
When is it reasonable to stop investigating?
When the contamination has been absent for several times its previous recurrence interval, with the detecting controls still running at full frequency, and when a plausible mechanism has been identified rather than merely a correlation. Stopping earlier is understandable under time pressure but tends to produce a recurrence that is harder to investigate, because intervening changes have obscured the original pattern. If the source genuinely cannot be found, say so in the record, describe what was excluded, and keep the monitoring in place rather than declaring the matter closed.
What makes contamination tractable is treating it as an investigation with a written record rather than a cleaning problem with a deadline. Characterise before changing, map exposures on paper, design comparisons where one thing differs, keep the contaminated material until the end, and convert whatever is found into a control that does not depend on anybody remembering. It takes longer than opening fresh reagents, once. It takes far less time than opening fresh reagents every fortnight for a year.
