An untargeted proteomics study surveys many proteins rather than testing only a short predefined panel. Its results can guide follow-up questions after peptide exposure, but the size of the output table does not make it a complete inventory of the sample or a catalogue of established mechanisms.
Start with what entered the reported dataset
Find the sample type, collection time, preparation and measurement approach before reading the protein list. These define the experiment you are interpreting. Do not silently extend a result from the analysed fraction to every protein in the original biological system.
Separate the number of acquired signals, identified peptides, reported protein entries and entries retained for statistical testing. These are different totals, even when an abstract presents only one headline figure.
| Stage | Fictional count | What the count describes |
|---|---|---|
| Protein entries reported | 2,400 | Identification output |
| Entries passing quantitative filters | 1,800 | Analysed comparison set |
| Entries meeting change criteria | 120 | Selected result set |
The 120 selected entries are 6.7% of the 1,800 tested entries, approximately. They are not demonstrated to be 6.7% of all proteins in the biological system. The denominator belongs to the analysis.
If an entry is missing, look for whether it was never identified, lacked usable quantitative observations or was excluded by a filter. These states should not all be translated into biological absence.
Read identification confidence at the correct level
The original double-competition study on peptide-level error control distinguishes peptide-spectrum matches, peptide discoveries and protein discoveries. False-discovery assessment belongs to a specified level; a stated threshold should not be copied between levels without checking the analysis.Lin and colleagues — Improving peptide-level mass spectrometry analysis via double competition (opens in a new tab)
Read the methods for the search database, identification rules and reported confidence measures. An impressive spectrum-level statistic alone does not tell you how uncertainty was handled when evidence was assembled into protein entries.
Also inspect the names used for rows. Where an output reports a protein group, preserve that grouping in your summary until the paper provides evidence for a more specific assignment.
Identification confidence and evidence for a treatment-associated difference are separate issues. A securely identified protein may show no convincing change, while a numerical contrast is not useful if its molecular assignment is unresolved.
Check how peptide signals became a protein comparison
Cox and colleagues’ MaxLFQ study describes an approach that combines peptide ratios and normalisation to estimate relative protein abundance. It illustrates why a protein-level quantitative output is a calculated result rather than simply one raw peak height.Cox and colleagues — Accurate Proteome-wide Label-free Quantification by Delayed Normalization and Maximal Peptide Ratio Extraction, Termed MaxLFQ (opens in a new tab)
Identify the actual method used in the paper. A label-free intensity, a labelled-channel ratio and an absolute calibrated amount should not be given the same interpretation merely because they appear beside a protein name.
For a fictional result labelled 1.8-fold relative abundance, retain the comparison group and direction. That value does not by itself mean 1.8 molecules per cell, an 80% increase in enzymatic activity or the same change in every sample.
Read the biological replicates and handling of unavailable values. If missing observations were replaced computationally, distinguish the modelled inputs from directly observed measurements when assessing the result.
Use the selected proteins to frame follow-up questions
A discovery list helps identify patterns worth investigating. Ask whether the reported effect sizes, uncertainty and selection criteria support the specific candidates highlighted in the discussion.
A pathway name attached to several selected proteins does not itself establish that the whole pathway changed activity. Follow how that interpretation was generated and whether a separate functional experiment addressed it.
Avoid selecting one attractive protein from a long table and presenting it as the only finding. State its place in the wider analysis and whether the paper treats it as exploratory or confirms it independently.
A useful evidence summary gives the sampled material, quantitative basis, analysed denominator and supported protein-level contrast. It then describes the proposed biological explanation with the additional evidence, or uncertainty, that accompanies it.
Sources and further detail
- Lin and colleagues — Improving peptide-level mass spectrometry analysis via double competition (opens in a new tab)
Original methods-paper introduction and error-control discussion read for the distinction between PSM, peptide and protein discoveries. No algorithm implementation or universal threshold prescribed.
- Cox and colleagues — Accurate Proteome-wide Label-free Quantification by Delayed Normalization and Maximal Peptide Ratio Extraction, Termed MaxLFQ (opens in a new tab)
Original methods-paper indexed passages read on combining peptide ratios and normalisation for relative protein quantification. No historical missing-value approach adopted as a general recommendation. Dataset counts are original.
Sources checked 19 September 2026. Worked examples are illustrative unless a supplied report is explicitly identified. This article has not undergone independent scientific peer review.