Sizing a band off a gel is not a lookup. Migration distance through agarose is roughly linear in the log of fragment size, which is why a ladder exists at all: it gives you a set of known points to fit a line through, and once the line is fitted you can read anything off it. The catch is that the fit has to be good, and the R² on it tells you whether it is before you trust the answer.
Sizing a band off a gel is not a lookup. Migration distance is roughly linear in the log of fragment size, so you fit a line to your ladder first and then read the unknowns off it. The R² on that fit tells you whether your ladder is any use before you trust the answer.
Use the same units for every distance you type — millimetres from the bottom of the gel, or centimetres, it does not matter as long as it is consistent. The fit is log10(size) against distance, so a good ladder gives a straight line with a negative slope. At least three bands are needed.
Why it is a log scale
DNA is negatively charged and migrates towards the anode at a rate set by how much it can move through the agarose mesh. Small fragments wriggle through the pores easily; large ones barely move at all. The relationship is not linear, and over the range of a single gel it is well approximated by:
log10(fragment size in bp) = m × (migration distance) + b
where m is the slope and b the intercept. So you fit a straight line to log(size) against distance, and the slope comes out negative, because bigger fragments travel less far.
This is the single most useful thing to internalise: largest band, nearest the wells. If a fitted slope ever comes out positive, either your distances are measured from the wrong end of the gel or your ladder rows are the wrong way up.
How to measure the distance
Measure from the bottom of the gel, or from the wells if you are consistent, but pick one origin and never mix them. Millimetres and centimetres both work as long as every number in the calculation uses the same unit, because the slope absorbs the scale factor.
Use the centre of each band, not an edge. Overloaded bands are wider, and the leading edge migrates further than the centre, so using the front of a fat band biases your answer towards an under-estimate of size. If a band is saturated, it has run at the wrong intensity and its apparent position is not usable at all.
Use a ruler or a gel imager rather than judging by eye against the gel tank. Most of the error in this procedure is measurement error, not chemistry.
Reading the fit
You need at least three bands to fit a line, and more is much better. Four to six well-spaced bands across the range you care about is the practical minimum.
R² is the number to look at. It is the proportion of the variation in log(size) that your line explains. Anything above about 0.95 is fine for routine sizing, which is why this tool marks anything below that as unusable rather than reporting a number it does not trust.
A poor fit has a small number of causes, and they are worth being able to name:
- Distances measured from a different origin than you think, or some rows measured one way and some the other.
- Too few bands, or all your bands bunched at one end of the gel so the line is only constrained over a narrow range.
- A band that is saturated and running at the wrong intensity.
- A gel that has not finished running, or has run off the end for the smallest bands.
Extrapolation is the quiet trap
Your fitted line is trustworthy between the ladder bands that bracket it. Outside that range you are extrapolating, and the line will happily give you a confident number with nothing behind it. A sample that appears larger than your biggest ladder band has not been measured, it has been guessed.
It is also worth remembering that a gel is a rough instrument. Real PCR products and restriction fragments often differ from their predicted size by 5–10% simply because the migration is not perfectly idealised. Treat a band size as one or two significant figures unless you have unusually good data.
Worked example
Take a ladder with bands at distances 10, 20, 30, 40, 50, 60, 70 and 80 mm, sized 10000, 7499, 5623, 4217, 3162, 2371, 1778 and 1334 bp. That is a log-linear ladder, and fitting it gives an R² of essentially 1.
Now an unknown band at 25 mm. Reading it off the fitted line gives roughly 7200 bp, and because 25 mm falls between the 20 and 30 mm bands, it is bracketed and the answer is sound.
Feed 30 mm back in as though it were unknown and you get 5623 bp, its own ladder value. That round trip is the check worth doing on your own data: take a ladder band, treat it as unknown, and confirm the tool gives you back the number printed on the tube. If it does not, your measurements or your units are inconsistent before you go near a real sample.
What this cannot do
This fits a straight line, which assumes one gel under one set of conditions. It cannot resolve bands that are closer together than your measurement precision, and it cannot correct for a distorted gel, where bands smile or frown because the gel was not level or was not cast properly. For accurate sizing of large fragments, or anything where the exact size matters commercially, use a capillary system or a pulsed-field gel, where the relationship between size and migration is set by the instrument rather than by the mesh.
Frequently Asked Questions
How do you size a band from a gel?
You fit a line to log of fragment size against migration distance using your ladder bands, then read the unknown off that line. It is not a lookup against a printed table, because the relationship depends on the gel, the buffer and the run time.
Why is the slope negative?
Because bigger fragments migrate less far. A positive slope almost always means your distances were measured from the wrong end of the gel, or the ladder rows are the wrong way up.
What R-squared do I need to trust the fit?
Anything above about 0.95 is fine for routine sizing. Below that the fit will not give reliable sizes, and the usual causes are too few bands, mixed measurement origins, or a saturated band running at the wrong intensity.
How many ladder bands do I need?
Three at the absolute minimum to fit a line, but four to six well spaced across the range you care about is the practical minimum. All your bands bunched at one end only constrains the fit over a narrow range.
Can I size a band bigger than my largest ladder band?
You can put a number on it, but that number is extrapolation with nothing behind it. Your fitted line is only trustworthy between the bands that bracket it, so anything outside the ladder range has not really been measured.
Should I measure to the centre or the edge of a band?
The centre. An overloaded band is wider and its leading edge runs further than the centre, so measuring to the front of a fat band biases you towards under-estimating the size.
How accurate is gel sizing?
Roughly 5 to 10 per cent for well-run routine gels, mostly limited by measurement rather than chemistry. Treat the result as one or two significant figures unless you have unusually good data.