The Mass on Your Certificate Was Calculated, Not Weighed

Deconvolution hands back a single tidy number with a decimal point on it, and that number is an inference. The algorithm cannot tell you when its assumptions failed. It returns the same confident figure either way.

Four ways the reconstruction goes wrong

Start with the failures, because they explain the machinery better than the machinery explains itself. Where a preparation holds both the target peptide and something closely related, the two ion series interleave. The software may then average them into one mass sitting between the real ones, or divide them along the wrong lines, and which of those happens depends on the algorithm and the parameters it was handed.

Sodium and potassium attach readily, putting 22 and 38 daltons onto the neutral molecule and landing at offsets a practiced reader learns to recognize. Software that does not model them can emit peaks in the reconstruction that are pure artifact yet look exactly like genuine related species.

A scan window that clips one end of the charge distribution leaves fewer observations to pin the answer down. Uncertainty grows, and nothing in the output says so.

Smoothing set too aggressively buys a tidy-looking reconstruction by erasing minor components that were really there.

Why the axis is not a mass axis

Protons are what electrospray adds. Basic sites on a peptide accept them: lysines, arginines, histidines, the N-terminus. Across the population of molecules entering the source, different numbers get picked up, so one compound arrives as a family of ions.

What the analyzer sorts is mass-to-charge. An ion of neutral mass M bearing n protons appears where the quantity M plus 1.00728n has been divided by n. One peak per value of n, and the collection of them is the charge envelope.

Solving it with two peaks and a pencil

Any two neighboring members of the envelope are sufficient, because the charge differs by exactly one between them.

Suppose peaks at m/z 1123.6 and 843.0. Call the first n-plus and the second one charge higher. Each implies a neutral mass: strip n proton masses off n times 1123.6, strip one more proton mass off (n+1) times 843.0, and set the two expressions equal. That resolves to n = 3, putting the neutral mass close to 3367.9 daltons. Every remaining peak should now fall where 3367.9 says it should, and that agreement is what confirms the assignment.

Software performs the same operation over the entire envelope simultaneously and outputs one reconstructed trace plotted against neutral mass. The certificate figure normally comes from that trace.

So the mass was derived, not read. An algorithm produced it from measurements while assuming the peaks belong to a single species, that the gaps between them are protons, and that enough of the envelope was captured to fix the answer. Hold those assumptions and the result beats any individual measurement, since it averages several. Break them and the output still arrives, indistinguishable in appearance from a correct one.

Reading charge off the isotopes

Position within the envelope is the weaker route. Carbon-13 gives every organic molecule an isotope cluster, and those component peaks sit about 1.0033 daltons apart in mass, which on the m/z axis becomes 1.0033 divided by the charge.

No assumptions are needed for this and the answer is unambiguous. What is needed is resolving power sufficient to see the individual isotope peaks at all. Instruments without it fall back on gaps between whole charge states to assign charge.

Two masses, roughly two daltons apart

Whether the isotope cluster was resolved also decides which mass gets reported. Resolve it and monoisotopic mass is available, built from the lightest isotope of each element. Fail to resolve it and the reconstruction yields an average mass, weighted over the natural isotope distribution.

The gap between the two runs at roughly a dalton for every thousand, so on a 3,000 dalton peptide they sit about two daltons apart. Check a reported value against the wrong calculated one and the discrepancy that appears is precisely that. Monoisotopic versus average mass covers the distinction.

What belongs next to the number

The raw spectrum, printed beside the deconvoluted one. It shows the envelope, lets a reader confirm the charge states are spaced sensibly, and exposes adducts and second species the reconstruction may have swallowed.

Handed a deconvoluted mass by itself, with no raw trace and no note of whether it is monoisotopic or average, a reader has one number standing in for a calculation nobody can audit. Usually correct. Never checkable.

Worth knowing, then, that a bare figure on a certificate is the output of a model with settings rather than something an instrument read off directly, and that the document ought to state which kind of mass it is. Related reading: MALDI versus ESI and mass accuracy in parts per million.