The first gel most people run is interpreted as a yes or no question. Is there a band where the band should be? If yes, the reaction worked. If no, it did not, and the usual next step is to repeat it with slightly different conditions and hope.
A gel is considerably more informative than that. Every visible feature of a lane is caused by something specific. A band sitting higher than expected, a band that is fuzzy rather than sharp, a smear running the length of the lane, a bright halo at the well, a lane that curves towards its neighbour: each of these has a small set of plausible causes, and learning to read them turns a pass or fail readout into a diagnostic tool that tells you what to change.
This piece treats the gel as a report rather than a verdict. The physics comes first, because almost every interpretive rule follows from it, and then the vocabulary of common defects and what each one indicates.
Key takeaways
- Migration distance depends on molecular size because the gel matrix impedes larger molecules more, not because charge differs.
- Matrix choice sets the resolving range: agarose for large nucleic acids, polyacrylamide for small ones and for proteins.
- A ladder is a calibration curve, and size estimates are only reliable within the range where its bands are well separated.
- Band intensity supports rough relative quantification only, and saturates readily on a bright detector.
- Most defects, including smearing and distorted bands, are caused by loading, sample quality or running conditions rather than by the reaction under test.
How an Electric Field Sorts by Size
Nucleic acids carry a negative charge on every phosphate in the backbone, and that charge is uniformly distributed along the molecule. A longer fragment has proportionally more charge and proportionally more mass, so in free solution the two effects cancel almost exactly and fragments of different lengths move at nearly identical speeds. Free electrophoresis therefore separates nucleic acids very poorly.
The gel is what creates the separation. It is a mesh of polymer strands with pores of a characteristic size distribution, and a molecule migrating under the electric field must find a route through it. Small molecules thread through easily and travel close to their free-solution speed. Larger molecules must reptate, worming their way end-first through the mesh, and this becomes progressively harder as size increases. Migration rate therefore falls as size rises, and it does so approximately logarithmically over the useful range, which is why size markers are plotted against the logarithm of length.
Two limits bound the useful range. Below a certain size, all fragments pass through the pores so easily that they separate poorly and run near the dye front. Above a certain size, all fragments struggle equally and pile up near the well. The gel resolves well only between those extremes, and choosing the matrix is essentially choosing where that window sits.
Proteins behave differently and require an extra step. They carry mixed charges determined by their amino acid composition and pH, and they fold into compact shapes of varying compactness. To sort them by size, the method must first destroy both variables. A denaturing detergent unfolds the protein and coats it with a roughly uniform negative charge per unit length, and a reducing agent breaks disulfide bonds so the chain is genuinely linear. Only after that treatment does migration track size in the way it does for nucleic acids. A protein gel run without full denaturation is measuring something else entirely, which is occasionally the intention and more often a mistake.
Agarose Versus Polyacrylamide Matrices
Agarose is a polysaccharide that forms a gel with large pores when a hot solution cools. Its pore size falls as concentration rises, so the same material covers a wide range simply by changing how much is dissolved. It is easy to prepare, non-toxic, and it casts into a horizontal tray with a comb, which is why it dominates routine nucleic acid work.
Polyacrylamide forms by chemical polymerisation of acrylamide monomer with a cross-linker, and it produces a much finer, more uniform mesh. It resolves small differences in size that agarose cannot, down to single-base differences under the right conditions. It is cast between glass plates and run vertically, takes longer to prepare, and the unpolymerised monomer is a neurotoxin that requires careful handling.
| Property | Agarose | Polyacrylamide |
|---|---|---|
| Pore size | Large, tuned by concentration | Small, tuned by monomer and cross-linker |
| Typical use | Nucleic acid fragments from hundreds to many thousands of bases | Small nucleic acids, sequencing-scale resolution, all protein work |
| Resolution | Modest, adequate for distinguishing clearly different sizes | High, can separate near-identical lengths |
| Preparation | Melt and pour, minutes | Polymerise between plates, longer |
| Format | Horizontal, submerged | Vertical, buffer tanks |
| Hazard | Low, aside from the stain used | Unpolymerised monomer is toxic |
The practical selection rule follows the size of what you expect. If the fragments of interest differ by a large proportion of their length, agarose at an appropriate concentration will resolve them. If they differ by a small proportion, or if you are working with anything below a few hundred bases, polyacrylamide is the right choice and no amount of adjusting agarose concentration will substitute.
Within agarose, concentration is the main lever. A low percentage gel has large pores and spreads out large fragments while letting small ones run off the end. A high percentage gel resolves small fragments and compresses everything large into an unresolved band near the well. Running the wrong concentration is the most common reason a gel appears to show one band when the sample contains several.
Ladders and Estimating Fragment Size
A ladder is a mixture of fragments of known length, and it functions as a calibration curve run in parallel with the samples. Because migration depends on gel concentration, buffer, voltage, temperature and run time, no absolute relationship between distance and size exists. The ladder converts a distance measured on this gel into a size estimate.
Reading it well means respecting the shape of the calibration. The relationship between migration distance and the logarithm of size is roughly linear only in the middle of the range. Near the top of the gel, large fragments are compressed together and small differences in position correspond to enormous differences in size, so estimates there are poor. Near the dye front, resolution collapses again. A band sitting between two well-separated ladder bands in the middle of the gel can be sized with reasonable confidence; a band above the highest ladder band can only be described as larger than it.
Ladder choice therefore matters as much as ladder presence. A ladder whose bands bracket the expected product tightly gives a much better estimate than one that happens to be in the freezer. Where the exact size matters, running the ladder in more than one lane, on both sides of the samples, corrects for the slight differences in migration across the width of a gel, which are real and are caused by uneven buffer depth, temperature gradients and small variations in gel thickness.
It is worth remembering what a size estimate from a gel actually is. It is an estimate of apparent size under these conditions, and several things can make a molecule run anomalously. Circular and supercoiled DNA migrates differently from linear DNA of identical length, so an uncut plasmid preparation typically shows multiple bands that do not correspond to different amounts of material. Strongly bent or unusually sequenced fragments can run slightly off. In protein gels, heavy glycosylation makes a protein run larger than its polypeptide mass. A gel measures mobility; size is the inference.
Band Intensity and Rough Quantification
Brightness in a stained gel is a function of how much stain is bound, which is a function of how much material is present, but the relationship is neither linear nor identical across bands.
Intercalating stains bind along the length of a nucleic acid, so a long fragment binds more stain per molecule than a short one. Two bands containing equal numbers of molecules will therefore differ in brightness in proportion to their lengths. Comparing brightness between bands of very different size without accounting for this leads to significant misreading, and it is the reason a faint high-molecular-weight band can represent less material than its appearance suggests, while a bright small band may represent less than it appears.
Detector saturation is the second limitation. Modern imaging systems have a limited dynamic range, and a band bright enough to saturate the sensor records no additional signal as more material is added. Two lanes that both saturate look identical regardless of whether one contains twice the material. Any attempt at comparison requires an exposure in which no band of interest is saturated, which usually means taking several exposures and using the dimmest usable one.
Within those limits, relative comparison of similar-sized bands on the same gel is legitimate and useful. Judging whether a digest went to completion, whether one sample contains noticeably more product than another, or whether a loading control is consistent are all reasonable uses. What a gel does not support is precise quantification, and where an actual concentration is needed, a fluorometric or spectrophotometric measurement of the sample is the appropriate tool.
Smearing, Streaking and Their Causes
A smear is a continuous distribution of material rather than a discrete band, and it means the sample contained a continuous distribution of sizes or that something disrupted migration.
Degraded nucleic acid is the classic cause and produces a characteristic low, diffuse smear as the fragments have been randomly cleaved into a continuum of shorter pieces. If the sample also lacks the sharp bands it should have, degradation is the leading explanation, and the source is usually nuclease contamination or repeated freeze and thaw cycles.
Overloading produces a different picture: a bright band with a trailing smear beneath it and often a distorted, curved shape. Too much material in one well exceeds the matrix’s capacity to keep it in a narrow zone, and the excess trails behind. The fix is to load less, which people resist because the band was clearly visible, but a photograph of an overloaded lane cannot be interpreted quantitatively at all.
Streaking upward from the well, or material trapped in the well itself, points to something in the sample that will not enter the gel. Undigested protein, high salt, residual detergent or precipitated material all do this. Where a sample refuses to leave the well entirely while its neighbours run normally, sample cleanup rather than gel conditions is the issue.
Non-specific amplification in a reaction under test also produces a smear, and it is important to distinguish it from a technical smear. The signature is a smear that appears in the reaction lanes but not in a control lane run from the same stock, and it usually coexists with the expected band rather than replacing it. Running the appropriate controls is the only way to make that distinction confidently, which is the argument for including them even when the gel is expected to be routine.
Loading Errors and Well Damage
A surprising fraction of ambiguous gels are ambiguous because of what happened in the first thirty seconds.
Punctured wells are the most common physical fault. A pipette tip pushed through the thin floor of a well allows sample to leak beneath the gel, where it runs as a faint diffuse shadow or vanishes entirely. The lane looks empty or weak and the reaction gets blamed. Loading with the tip just above the well and letting the dense loading buffer carry the sample down avoids this entirely.
Loading buffer itself does two jobs and both matter. The density agent makes the sample sink rather than diffuse into the running buffer, and the tracking dyes provide a visual front so the run can be stopped before the sample of interest runs off. Omitting it, or adding too little, produces weak lanes with material lost into the buffer. Adding far too much can distort migration in the lane.
A bubble in a well displaces sample and can deflect the band, producing a lane that appears to have run at an angle. A comb removed roughly tears the well walls, and a torn wall lets adjacent lanes bleed into one another, which is the origin of many apparent contamination panics.
Uneven running conditions distort whole gels rather than single lanes. If the buffer level is too low, the gel dries and heats unevenly and lanes curve. If the voltage is too high, the gel heats, resolution degrades, and bands become fuzzy. Running slower and cooler is almost always better than running fast, and the time saved by a high voltage is regularly lost to repeating the gel.
Documenting Gels for Publication
An image intended for a record or a publication carries obligations that a quick check on a screen does not.
Capture the raw image at the highest bit depth the system offers and keep it. Adjustments made for presentation should be made on a copy, and the original file should remain available, because it is the only evidence of what the detector actually recorded.
Adjustments to brightness and contrast are acceptable when they are linear and applied uniformly to the entire image, including the ladder and any background. What is not acceptable is adjusting part of an image, applying non-linear corrections that selectively suppress background or enhance faint bands, or erasing a blemish. Journals increasingly examine submitted images for evidence of local manipulation, and the tools for detecting it are good.
Cropping is permitted and often sensible, but the crop must not hide relevant context. A cropped lane presented without its neighbours conceals whether the sample smeared, whether adjacent lanes bled, and whether the ladder was in range. Where lanes from different parts of a gel are placed side by side, that rearrangement must be shown with a clear dividing line and stated in the legend. Splicing lanes together without indication is a serious integrity problem regardless of intent.
The legend should let a reader reconstruct the experiment: matrix and concentration, ladder used, stain, what each lane contains, and which controls were included. An image of a gel with unlabelled lanes and no ladder is not evidence of anything, however convincing the band looks.
Frequently asked questions
Why does my band appear at the wrong size when the reaction should be specific?
Several explanations compete. Anomalous migration is common with structured or circular molecules, so an uncut plasmid or a product with strong secondary structure can run well away from its expected position. The gel concentration may place the fragment outside the resolving range, compressing it against neighbours. The ladder may be degraded, which shifts the calibration rather than the sample. And the product may genuinely be the wrong thing. Running a fresh ladder alongside a known control at a suitable gel concentration usually separates these possibilities quickly.
How much sample should I load?
Enough to see clearly and no more, which in practice means starting well below what feels sufficient and increasing if needed. Overloading is more damaging than underloading, because it distorts band shape, generates trailing smears, saturates the detector and can obscure a nearby band entirely. If a faint band must be visualised, a longer exposure on a sensitive imager is a better route than more material in the well.
Can I reuse running buffer?
It works for a limited number of runs and then stops working in a way that is easy to misattribute. Electrophoresis consumes buffering capacity and shifts the pH at the electrodes, so exhausted buffer produces slow, distorted runs with fuzzy bands and excessive heating. Since buffer is inexpensive relative to a repeated experiment, the sensible policy is to replace it on a fixed schedule rather than when problems appear.
What does it mean when a band is fuzzy rather than sharp?
Fuzziness generally indicates that molecules of the same size did not stay in a tight zone during the run. Excessive voltage and the resulting heat is the most frequent cause, followed by exhausted buffer, a gel that was not fully set before loading, and a sample loaded in a volume too large for the well. In protein gels, incomplete denaturation produces the same appearance, because a population of partially folded molecules migrates over a range of apparent sizes.
Is a gel still worth running when a more quantitative method is available?
Frequently, yes, because the gel answers a different question. A fluorometric measurement tells you how much nucleic acid is present but nothing about whether it is intact, whether it is one species or several, or whether an unwanted product came along. The gel provides that structural picture cheaply and quickly. The two are complements, and the sensible workflow uses the gel to confirm identity and integrity and a dedicated method to establish concentration.
The habit worth building is to look at a gel before deciding what it means. Note where the ladder resolves and where it does not, check whether any band is saturated, look at the wells, look at the shape of the lanes, and only then ask whether the expected band is present. Nearly every gel that gets described as inconclusive is in fact telling you precisely what went wrong, in a vocabulary that takes a term or two to learn and then never stops being useful.
