Troubleshooting sparse identifications

A short identification list is a symptom with a short decision tree. It is not a verdict that the biology was empty. Work from the tube toward the search, and stop at the first broken step. The vocabulary you need is small and exact: total ion current, base peak, peptide-spectrum match, and protein group. How a passed match becomes a protein name is the subject of bottom-up proteomics in plain language. How the peptides should have been made is preparing peptides for mass spectrometry.

Four words that keep the tree honest

Total ion current, often shortened to TIC, is the summed signal of the ions recorded in a scan, plotted across the chromatographic time. A TIC that is flat and near the solvent background means the instrument saw little that could be fragmented. A TIC that is a broad hump can mean the column was overloaded with polymer or salt rather than with peptides.

The base peak is the most intense ion in a single scan. A base-peak chromatogram draws that intensity over time. It is harsher than the TIC: one dominant polymer series can look "busy" on a TIC and still show a base peak that marches in regular mass steps, which is the signature of a detergent or a polyethylene glycol, not of a tryptic peptide.

A peptide-spectrum match, a PSM, is one fragmentation spectrum accepted as one peptide sequence at a stated score. Sparse identifications means few PSMs survived the threshold, or few of those PSMs assembled into protein groups. A protein group is the set of accessions that share the observed peptides and cannot be separated. Ten PSMs that all belong to keratin are one disappointing group, not ten proteins of interest.

Walk the tree in order

First ask whether peptides were present after cleanup. A UV trace at peptide-typical wavelengths, or a colorimetric peptide assay used the way its maker describes, answers a quantity question. If that reading is at the blank, the digest, the cleanup or the upstream protein amount failed. Do not retune the mass spectrometer to compensate for an empty vial. How laboratories estimate protein concentration is the check before the enzyme ever goes in.

If peptide was measured and the column still shows no peaks, the failure moved. The autosampler may have injected air, the trap may have been blocked, or the peptides may have stuck to a tube that was never wetted with a solvent the protocol allows. A solvent blank injected immediately afterwards tells you whether the column and the source are alive. A blank that looks identical to the sample means the sample never arrived.

If the TIC or the base-peak trace shows signal, decide what the signal is. Peptide MS1 scans tend to show peptide isotope envelopes and several charge states across a reversed-phase gradient. Detergent and polymer look like repeating mass ladders and broad late or early humps. A file full of those humps should be rescued at the bench, with a cleanup class that removes the surfactant you used, not by widening the search.

Only then open the search. Confirm the taxon. A mouse lysate searched as human can return a thin list of lucky homologs or a list you do not trust. Confirm the cysteine modification. If the proteins were alkylated with iodoacetamide and the search left cysteine unmodified, most cysteine peptides miss. If the search demanded carbamidomethyl-cysteine and the tube was never alkylated, the same peptides miss. UniProt is a usual source of the sequences; the version has to be the one the search used.

Then look at the false-discovery threshold on that particular file. The rate is a property of the accepted list, judged with decoy sequences. A tiny file has a coarse decoy count. A threshold copied from a deep fractionated study can reject every match in a gel band. Record the threshold, the number of candidate spectra and the number of PSMs. If loosening it within a pre-agreed pilot range fills the list with decoys, the spectra were not peptide spectra.

Last, read the protein groups you did get. If they are keratin, trypsin autolysis and the antibody or the bead protein from a pull-down, the instrument and the search worked. The sample of interest did not compete. That is a handling and loading conclusion.

Amount, column and search settings as classes

Peptide amount is a class of measurement, not a single brand of assay. Chromatographic signal is a UV detector or the mass spectrometer itself. Search settings that move a sparse list most often are taxon, cysteine fixed modification, enzyme, and the false-discovery cut. Variable modifications are a poor rescue. Adding every side reaction because the list is short inflates the search space and can make the decoy problem worse.

A public peptide resource such as PeptideAtlas can tell you whether a protein you expected is even known to yield observable peptides. It cannot repair a file that never contained those peptides. Method collections such as protocols.io are where cleanup variants are written down. Follow the protocol you ran, including its failure branches, rather than a neighbouring protocol with a different detergent.

Where you stop and change the experiment

If the peptide assay is blank, stop. Remake the digest from a reserved aliquot, or go back to lysis. Re-searching will not invent ions.

If the peptide assay is fine and the column is blank, stop. Check injection, trap and vial. A second aliquot injected on a column that just ran a known standard tells you whether the standard still appears. If the standard is also gone, the instrument method is the branch, and the samples wait.

If the trace is polymeric, stop. Change or repeat cleanup. Do not add the polymer's mass as a modification and hope the proteins appear underneath.

If the trace looks like peptides and the search is empty, change one search setting that you can justify, on a copy of the analysis. Taxon or cysteine chemistry first. Compare PSM counts. Keep the setting that the chemistry supports, not the setting that merely lengthens the list.

When the first cause is wrong

A healthy peptide number and a healthy standard can still hide a gradient that collapsed into a few minutes because a pump fault shortened the method. The TIC then looks like a spike rather than a hump, and the search returns a handful of abundant proteins. Compare the recorded gradient with the method you think you ran.

A correct taxon can still be the wrong database slice. A cultured human cell line searched against a mammal-wide redundant file can bury true matches in shared peptides and decoys. Narrow to the taxon you infected or transfected, and add a custom sequence if you expressed a fusion. That choice is a database decision, not a reason to turn the false-discovery threshold off.

Keratin plus a real but small list is mixed. Keep the keratin flagged, and judge the other groups by unique peptides and by whether they appear in a process blank. A process blank that contains the same "hits" ends the biological claim.

What a sparse list must not be asked to do

Do not convert a handful of PSMs into a pathway story. Do not report a research gel band as a clinical marker. Biosafety for the material you digested is an institutional decision; a short list does not make an infectious sample harmless, and a long list would not either. The Human Proteome Organization discusses how protein evidence is argued in research. It does not turn a troubleshooting file into a diagnostic report.

Heat, a stopped gradient, and a warm tube

Peptides left warm, especially in urea that was not kept cool, pick up chemical modifications and can also adsorb or dry down in a way the resuspension protocol did not expect. In a humid room a tube that was "dry" may still hold residual solvent. Resuspend as the cleanup protocol says and inject, or freeze. A power cut in the middle of a gradient writes a chromatogram that stops and, if the queue restarts blindly, a second partial run. Mark that file as failed chromatography. Do not average it with a completed replicate and call the biology variable.

What to send with a troubleshooting question

Send the peptide amount class after cleanup, a description of the TIC or base-peak shape, the taxon, the cysteine chemistry, the false-discovery thresholds, and whether a process blank was keratin. Say whether a known standard still ran. Those facts decide whether the next step is a new digest, a new cleanup, or a new search.

A discussion of scope can start from the shotgun discovery proteomics reference, the protein identification by LC-MS/MS reference, or the differential abundance reference, depending on whether you needed names or a comparison. Use the quote request and include the observation from the table that matches your file. That is a question about method. It is not a statement that a rerun has been booked.