Proteomic Tables: Peptide Evidence and Protein Abundance

From Protein Molecules to Table Entries

A proteomic table may contain a number for every protein in every sample, but that number is usually the end of a measurement-and-inference chain. In mass spectrometry, the instrument records signals from ions, often after proteins have been digested into peptides. In an affinity assay, a binding reagent recognizes a protein or an accessible part of it, and the assay records a signal related to that binding. In both cases, a measured signal is evidence used to estimate abundance; it is not automatically the number of protein molecules.

The distinction matters because the path from molecules to signal depends on the assay. Different peptides from the same protein can produce different ion signals, and different proteins can bind an assay reagent with different strengths. A table entry labelled “protein abundance” should therefore be read as an estimate with a measurement basis, not as a direct count unless the assay and calibration justify that interpretation.

Mass-Spectrometry Evidence and Protein Inference

In a common mass-spectrometry workflow, proteins are broken into peptides, the peptides are separated, and the instrument measures ions by mass-to-charge ratio. In tandem mass spectrometry, selected ions are fragmented; the resulting fragment pattern can be compared with predicted patterns for candidate peptide sequences. A reported peptide identification is therefore an inference from measured spectra, not a direct reading of a protein name from the instrument.

The identification step and the abundance step answer different questions. Identification asks which peptide sequence is sufficiently supported by the observed spectrum. Quantification asks how much signal is associated with that peptide in a sample. Protein inference then uses identified peptides to decide which protein or proteins are supported, and protein-level quantification summarizes peptide measurements assigned to those proteins.

A peptide sequence may be unique to one protein, or it may occur in several proteins. A unique peptide can support the presence of its assigned protein more specifically. A shared peptide supports the group of proteins to which it maps, but by itself cannot distinguish among them. This can arise for closely related proteins or protein isoforms. Thus, a protein table may report a protein group or a protein-level estimate whose evidence does not resolve every member separately.

Peptide intensity also varies for reasons beyond the amount of source protein. Digestion can release peptides with different efficiencies; peptides differ in solubility, ionization, and fragmentation; and instrument response can vary across the measurement range. Consequently, two peptides from equal amounts of different proteins need not have equal intensities. Even two peptides from the same protein need not have equal intensities.

A Simple Signal Model for Relative Abundance

A useful simplified model for a particular peptide \(j\) in sample \(i\) is

where \(I_{ij}\) is the measured peptide intensity, \(A_i\) is the amount of the source protein in sample \(i\), \(k_j\) is the response factor for peptide \(j\), \(q_{ij}\) represents sample- and processing-dependent factors such as peptide recovery, and \(\varepsilon_{ij}\) represents measurement error. This is a working approximation, most plausible over a range where response is roughly proportional to amount and the signal is not saturated.

If the same peptide is measured in two samples, and its response factor and recovery are sufficiently comparable, then a ratio can cancel the peptide-specific response factor:

This approximation requires more than using the same peptide label in both samples. It assumes comparable sample preparation and measurement, a usable signal in both samples, and no important sample-specific shift in recovery or interference. If these conditions fail, the intensity ratio may not represent the protein amount ratio. With several peptides, researchers may summarize peptide-level ratios, but that summary is still an estimate and depends on which peptides were identified, quantified, and assigned.

Building a Protein-Level Estimate from Peptides

When several peptides are assigned to one protein, a protein-level value may be obtained by summarizing their intensities or their sample-to-sample ratios. A simple strategy is to calculate a ratio for each peptide that is reliably quantified in both samples, then use a summary such as the median. This reduces dependence on any one peptide, but it does not remove the need for suitable identification, mapping, and comparability assumptions.

For example, if peptide ratios are \(1.0\), \(1.5\), and \(2.0\), the median ratio is \(1.5\). The interpretation is that the central peptide-level evidence supports an approximate 1.5-fold relative increase, provided the peptides all map appropriately and are measured comparably. A more elaborate pipeline may use different transformations, weighting, or model-based summaries. The value in the output table should be interpreted according to the method used to produce it.

Affinity-Based Measurements

Affinity-based assays use a reagent that binds a target protein, often an antibody or an aptamer. A label, enzyme reaction, or other detection system converts binding into a recorded signal. Depending on the platform, the signal may be fluorescence, luminescence, an electrical response, or another quantity. A table may report the raw signal, a normalized signal, a calibrated concentration, or a transformed score; these are not interchangeable.

For a simple calibration in a linear range, the signal can be approximated as \(Y=b+kC+\varepsilon\), where \(Y\) is observed signal, \(C\) is protein concentration, \(b\) is background signal, \(k\) is the calibration slope, and \(\varepsilon\) is measurement error. If standards establish \(b\) and \(k\), an observed signal can be converted to a concentration estimate. This interpretation depends on the sample signal lying in the calibration range and on the standards and sample behaving comparably.

Reading and Checking Proteomic Tables

Before interpreting a proteomic value, determine what level of evidence the row represents. A peptide-level table may include a sequence, spectrum or identification score, and a peptide intensity. A protein-level table may instead report an estimate assembled from multiple peptide measurements. An affinity-assay table may contain a raw or normalized response, or a concentration obtained from calibration. A label such as “abundance” does not by itself reveal which of these quantities is stored.

Useful checks focus on whether the evidence supports the stated level of interpretation. For mass spectrometry, ask whether peptide identifications meet the stated confidence criteria, whether the peptides map uniquely or are shared, and whether the protein estimate is based on multiple consistent peptides or a limited subset. For affinity assays, check the identity and specificity of the binding reagent, the calibration range, and whether the reported number is a signal or a calibrated concentration. These checks do not prove that an estimate is correct; they identify assumptions that matter to its interpretation.

A table can also hide differences in measurement procedures. For example, a protein-level value based on one peptide is not evidentially equivalent to a value supported by several peptides. Similarly, a calibrated concentration is not automatically comparable across assays that use different reagents or calibrations. The assay definition and processing steps are part of the meaning of the data.

Common Pitfalls, Alternatives, and Limitations

• Equating intensity with amount. A larger signal may be consistent with more protein, but response factors, recovery, interference, and calibration affect the relationship. State whether a number is an observed signal or an abundance estimate.
• Treating peptide identification as protein identification. A peptide can be shared by multiple proteins. Check whether the evidence distinguishes the particular protein or only supports a protein group.
• Assuming all peptides are equally informative. Peptides differ in detectability and response. A protein estimate based on one peptide can be more sensitive to that peptide’s behavior than a summary supported by several suitable peptides.
• Calling an affinity signal a concentration without calibration. A signal becomes a concentration estimate only through a calibration model with appropriate range and assay assumptions.

Mass spectrometry and affinity assays provide complementary routes to protein measurement. Mass spectrometry can provide sequence-based peptide evidence and can support estimates from multiple peptides, but peptide detection and protein inference can be difficult. Affinity assays can be sensitive and suited to measuring selected targets, but interpretation depends strongly on reagent specificity, binding behavior, and calibration. Neither method is universally superior: the measurement target and the evidence needed determine which is appropriate.

Both approaches have limitations. A mass-spectrometry signal can be affected by peptide-specific response and sample preparation, while peptide mapping may not resolve related proteins. Affinity signals can be affected by cross-reactivity, epitope accessibility, and matrix effects. In either case, the reported estimate is conditional on the assay and processing choices. It should not be presented as a direct causal or biological conclusion without additional evidence.