Insect protein has moved from novelty to mainstream. With Acheta domesticus (also known as the house cricket) now approved for human consumption in the EU and appearing in everyday products such as protein bars, the question is no longer whether insects belong in the food chain, but whether what is on the label is actually in the package. Sustainability claims, premium pricing and allergen safety all hinge on one thing: can you prove, at the species level, which organisms ended up in a finished, heavily processed food? Increasingly, that proof must come from analytics and liquid chromatography coupled to mass spectrometry (LC/MS) is emerging as the method of choice.

In a recent application note developed in collaboration with Agilent, our team tackled this particular challenge. We built an LC/MS peptide mapping workflow that reads the species “fingerprint” hidden in food, and applied it to five commercially available cricket-based protein bars (C-BAR01 to C-BAR05) as well as one insect-free control (N-BAR). The result: five peptides (unique to A. domesticus) were detected consistently in every cricket-containing bar and were entirely absent from the control. A clean, species-level confirmation drawn straight from the protein evidence, even in a matrix crowded with soy, rice, dairy, nuts and chocolate.

Why authenticating insect protein is hard

A finished protein bar is one of the least forgiving samples in food analysis. Its proteins arrive from many sources at once (soy, rice, whey, milk, peanuts, cocoa) and the cricket fraction can be as little as a few percent of the total. Manufacturing compounds the problem: grinding, heating and pressing degrade proteins and can mask or destroy the very markers an analyst is hoping to find. DNA-based tests, the traditional tool for species identification, also struggle here because processing fragments DNA and because the presence of a gene says little about whether the corresponding protein (and any allergen risk it carries) actually survived into the product. Two major problems in particular limit confident identification in this case:

• Shared peptides across related taxa: many tryptic peptides are common to entire insect families or classes, so they point to “an arthropod” or “a cricket relative” rather than to A. domesticus
• A crowded, processed background: species-specific peptides sit at low abundance against an overwhelming plant- and dairy-protein matrix, and data-dependent acquisition can only sample a fraction of everything that elutes.

How to read a cricket’s molecular fingerprint

Our workflow turns the bar’s protein content into a searchable set of peptides. Each sample is cryogenically ground, delipidated with hexane to strip out interfering fats, and its proteins are then extracted, reduced, alkylated and digested overnight with trypsin. The resulting peptide mixture is separated on an Agilent AdvanceBio Peptide Mapping C18 column over a 90 min gradient using the Agilent 1290 Infinity III Bio LC, and analyzed on an Agilent 6545XT AdvanceBio LC/Q-TOF in data-dependent acquisition mode: a biocompatible front end paired with a high-resolution, accurate-mass back end that is well suited to complex biological samples.

The measured MS/MS spectra are matched against curated protein databases (UniProtKB reference proteome, A. domesticus proteome and contaminant list) and then passed through Unipept, an open-source tool that maps each tryptic peptide to the narrowest taxon it can belong to. Peptides found in only one species become molecular fingerprints. Peptides shared across many taxa are flagged and set aside. It is this peptide-by-peptide bookkeeping that lets a single LC/MS run resolve “insect” all the way down to “house cricket”.

From thousands of peptides to five molecular IDs

The numbers convey the scale of the search. On average, each sample yielded on the order of 78,000 MS/MS spectra, of which roughly 8,750 were confidently matched to peptides and about 1,080 were unique. This corresponded to an identification rate of around 11 %, typical for a matrix this complex. Mapping those peptides taxonomically, every cricket bar returned 25-41 peptides from the Arthropoda phylum and 10-12 from the cricket subfamily Gryllinae. Crucially, five peptides were unique to A. domesticus itself (corresponding to three proteins). Those five sequences are the species-level signature, i.e. the molecular IDs that separate house cricket from any of its relatives.

Validation and complementarity

Two specific findings gave our result their credibility. First, the insect-free control bar returned zero peptides assignable to insect proteins. In other words, our approach did not render any false positives, while all five A. domesticus peptides appeared in every cricket-containing bar. Second, the signal intensity of those peptides tracked the declared cricket content across the range of formulations (from roughly 4.6 % to 13 % cricket powder), pointing towards a semi-quantitative reading rather than a simple yes/no call. The same workflow confirmed the rest of each label: it was able to distinguish cultivated soybean (Glycine max) from its wild relative Glycine soja, and picked out almond, peanut, oat and buckwheat exactly where the packaging said they should be.

There is a safety dimension too. The marker proteins are not arbitrary: tropomyosins are well-characterized invertebrate pan-allergens that are heat-stable and share IgE-binding epitopes with shellfish allergens. Apolipophorin-III has also been flagged in insect allergen studies. Detecting these proteins therefore performs a double duty: it not only confirms the species, it signals the presence of allergen-relevant material. An important side note here: while our workflow was able to detect specific allergen proteins, it does not measure their clinical allergenicity. The latter still requires dedicated immunological testing.

Implications for the novel food industry

For companies bringing insect-based foods to market, sequence-level protein evidence addresses several practical needs at once:

• Two-way label compliance: confirm which insect species is present, or confirm that insect protein is genuinely absent where a product claims to be insect-free.
• Allergen awareness: the ability to flag tropomyosin and apolipophorin-III in a single run, supporting allergen labeling and risk assessment.
• Built for processed products: because it reads peptides rather than intact DNA or whole protein, the approach works in heavily processed, multi-ingredient matrices where other methods lose resolution.
• A possible route to semi-quantitation: marker intensity that tracks cricket content opens a path to estimating how much insect protein a product actually contains, not just whether it is there.
• One method, full label: a single LC/MS experiment supports verification of multiple ingredient classes together (plant, animal and insect), supporting transparency across the entire formulation.

Conclusion

For food producers, ingredient suppliers and regulators navigating the fast-moving novel-food landscape, this work offers a ready-to-use blueprint for proving what is and what is not present in an insect-based product. As regulatory scrutiny and consumer expectations rise together, species-level, allergen-aware authentication is fast becoming a baseline requirement rather than a nice-to-have.

Do you want to have our application note all to yourself? You can download it here.

Are you bringing an insect- or novel-food product to market and need help with your analytics? Let’s talk before any label surprises bug you!