Every LabelCheck every label, every time.

← Cross-reactivity guides

Seed storage protein cross-reactivity

By Symon, Every Label ·

If you or your child reacts severely to a plant-seed food, the protein behind it usually comes from one family: the seed storage proteins. These are the proteins a plant packs into its seeds to feed the next generation, and they are built to survive, heat-stable and hard to digest, which is exactly why they drive the serious, whole-body reactions across tree nuts, peanut, sesame, soy, and the other legumes. Cooking, roasting, and baking do not defuse them. This is the protein hub for that whole group of foods.

There is one rule that matters more than any other here, and it leads the page on purpose: a positive test to a seed storage protein is a red flag for whole-body reactions, not a reassuring result. It is the opposite of the milder pollen-related proteins that often cause only an itchy mouth. The rest of this page explains the protein first, then sorts the seed foods by how strongly they actually cross-react, then tells you which ones only test positive without reacting, and then sends you to each food’s own page. Where a claim is a verified cross-reactivity fact, it is drawn from the project’s cross-reactivity floor. None of it replaces your allergist.

The molecular why: one family of proteins, packed into seeds

The seed storage proteins are the spine of this whole story, so they go first.

Every seed, a tree nut, a peanut, a sesame seed, a soybean, a lentil, has the same job: store enough protein to power the next plant until it can feed itself. The proteins that do that storage job fall into three related families, and they are the ones the immune system reacts to most often and most dangerously:

  • The 2S albumins (for example peanut Ara h 2, cashew Ana o 3, walnut Jug r 1, hazelnut Cor a 14, sesame Ses i 1, mustard Sin a 1). These are small, tough proteins and are the single most important markers of a serious, systemic allergy in this whole group.
  • The 7S vicilins (for example peanut Ara h 1, cashew Ana o 1, walnut Jug r 2, soy Gly m 5).
  • The 11S legumins (for example peanut Ara h 3, cashew Ana o 2, walnut Jug r 4, hazelnut Cor a 9, soy Gly m 6).

Two properties of these proteins drive everything below.

They are heat-stable and digestion-resistant. They are built to survive, so roasting, baking, salting, and cooking do not break them down, and they survive the stomach intact. That is why a reaction to a storage protein tends to be whole-body rather than staying in the mouth, and why it does not fade when the food is cooked.

How alike two seeds’ storage proteins are tracks how strongly the foods cross-react. When two seeds carry storage proteins that look alike to the immune system, an antibody trained on one can bind the other. The closer the proteins, the more an allergy to one predicts the other. Near-identical proteins (cashew and pistachio, walnut and pecan) mean a very high, predictable cross-reaction. More distantly related proteins (across different plant families) mean a weaker, less predictable one, often a positive test with no real reaction. The protein is what sets the distance, and the rest of the page is that map.

The rule this whole page exists to make: a positive storage-protein test is a red flag

This is the single most important point, and it is the opposite of what a “minor component” label might suggest. A positive test to a seed storage protein is a red flag for whole-body reactions, not a reassuring low-risk result. A positive result to a tree-nut storage protein (cashew Ana o 1 or Ana o 2, walnut Jug r 2 or Jug r 4, pistachio Pis v 2, Pis v 3 or Pis v 5, pecan Car i 2 or Car i 4, hazelnut Cor a 9 or Cor a 14) is a reason for strict avoidance and an epinephrine plan, confirmed with your allergist, never a reason to relax. These are 7S vicilins and 11S legumins, heat-stable and digestion-resistant, so roasting or cooking does not make the nut safe. If anyone reads a positive storage-protein result as “only a minor component,” that reading is wrong.

There is one documented exception within this protein family, and it shows how specific the biology is rather than softening the rule. In Brazil nut, a different storage protein, the 2S albumin Ber e 1, drives the severe reactions, and the 11S legumin Ber e 2 is genuinely minor. The lesson is not that Brazil nut is low-risk. It is that which storage protein lit up is what decides the danger, which is exactly what makes component testing worth doing.

The seed foods, grouped by how strongly they cross-react

This is the map: every part of the cluster, sorted by how much an allergy to one tells you about the others. The strength is the verified cross-reactivity record, not a guess. The tree-nut category map (the four groups, the label aisle) lives on the tree-nut cross-reactivity page; here the foods are sorted by the protein.

Very high: near-identical proteins, managed as a single allergy

At the top of the cluster sit two pairs whose storage proteins are so alike that allergists usually treat the two foods as one allergy.

  • Cashew and pistachio. They are close cousins (the Anacardiaceae plant family) and their storage proteins are near-identical: cashew Ana o 1 and pistachio Pis v 3 are homologous, and cashew Ana o 3 and pistachio Pis v 1 are near-identical 2S albumins. A person allergic to either is treated as allergic to both, and either is reintroduced only through a supervised challenge, never at home.
  • Walnut and pecan. They are the two most closely related tree nuts (the Juglandaceae family), and their storage proteins are near-identical: walnut Jug r 1 and pecan Car i 1 are homologous 2S albumins, and walnut Jug r 4 and pecan Car i 4 are homologous 11S legumins. Roughly 9 in 10 people allergic to one react to the other, and reactions can be severe. They are managed as one allergy the same way.

These are the part of the cluster where one allergy genuinely predicts the other, because the proteins are nearly the same molecule.

High: a real storage-protein axis, taken seriously and tested

Below the near-identical pairs are foods whose storage proteins are clearly related, not identical. The cross-reaction is real and documented rather than a test artifact, so it is taken more seriously than an unrelated food, but it is not an automatic shared diagnosis. These are tested, not assumed, and treated as a real question.

  • Hazelnut with walnut and pecan. Hazelnut is in a different plant family from walnut and pecan, but it shares the same class of storage proteins (its 2S albumins and the 11S legumin Cor a 9), so hazelnut, walnut, and pecan allergies frequently occur together, and patients sensitized to these proteins are at risk of severe reactions to both nuts. If one of these three is a known allergy, the others become a real question to test.
  • The legume triad: lentil, chickpea, and pea. Among the legumes, lentil, chickpea, and pea cross-react strongly through their shared storage proteins, and a child allergic to one frequently reacts to the other two on a supervised challenge. This triad behaves more like the tree-nut pairs than like the looser peanut-to-legume links below, so an allergy to one of the three is a reason to test the other two carefully with your allergist.
  • The two mustards. Yellow mustard and oriental (brown) mustard share their major storage-protein allergen, a near-identical 2S albumin (yellow mustard Sin a 1 and oriental mustard Bra j 1), so they cross-react. A mustard allergy generally covers both types.
  • Poppy seed and sesame. Poppy seed and sesame can cross-react in seed-allergic people. This is a documented seed-to-seed link to test rather than assume in either direction.

Cross-category: real cross-reactions that jump between plant families

The storage-protein machinery also links foods in different botanical families, where the cross-reaction is real enough to manage as avoidance even though the foods are not close cousins.

  • Fenugreek and peanut. Fenugreek is a legume used mainly as a spice, and peanut and fenugreek can cross-react, with fenugreek allergy often arising from a primary peanut allergy. If you are peanut-allergic, treat fenugreek (in curry blends, spice mixes, and some supplements) as a food to clear with your allergist rather than assume.
  • Lupin and peanut. Lupin is a legume whose flour is a common wheat substitute, and lupin allergy can develop in peanut-allergic people through cross-reactivity between their proteins. A peanut allergy is a reason to be cautious with lupin (often labeled “lupin flour”), not to assume tolerance. Lupin is not a food this page clears for anyone.

Tests positive, usually tolerated: where over-avoidance creeps in

This is the part most often gotten wrong, and it is the reverse of the storage-protein red flag above. Some foods in this group share enough storage protein to light up on a test, but the real-world reaction usually does not follow. A positive test here is a reason to test further, not a verdict to avoid. Each of these still routes through your allergist before you act on it.

  • Peanut and soy. Peanut and soy commonly show cross-sensitization on testing, but clinical cross-reactivity is rare and most people with peanut allergy tolerate soy. Confirm with your allergist before introducing it.
  • Chickpea and peanut. Chickpea and peanut commonly test co-positive, but clinically relevant cross-reactivity is much lower than that co-sensitization. Confirm with your allergist before introducing.
  • One legume allergy does not mean avoiding all legumes. Having one legume allergy, such as peanut, does not mean you must avoid every legume; most people with peanut allergy tolerate other legumes such as soy, peas, lentils, and chickpeas, because the cross-sensitization on testing is usually not clinically relevant. Confirm with your allergist before introducing any of them.

The honest line that ties this tier together: a positive panel across the cluster co-fires far more often than real reactions follow, so a broad positive result is a reason to test, not a list of foods to fear. The exceptions are exactly the very-high and high tiers above, where the proteins are close enough that the test and the reaction track together.

What is NOT in this cluster, and the corrections worth making

The cluster is broad, but it has hard edges. Two are worth stating plainly, and one points you to a different protein page entirely.

Coconut is a drupe, not a storage-protein seed. Coconut shares the word “nut” with the others but not the biology. It is botanically a fruit (a drupe) from the palm family, it does not carry the seed storage proteins that drive this cluster, and most people with tree-nut allergy tolerate it. Confirm with your allergist before introducing it rather than assuming it is automatically safe. Families often cut coconut on the strength of the name alone, and that is over-avoidance.

Almond does not behave like the rest of the storage-protein cluster. Almond carries the storage protein Pru du 6, but it is the least cross-reactive of the common tree nuts, and most people who test positive to almond tolerate it. The almond-to-hazelnut link people ask about most is not established on our verified floor, so this page does not assert it and does not deny it. The instruction is the relaxed one: a positive almond test, or another nut allergy, is not a reason to drop almond on its own; ask for component testing or a supervised challenge rather than removing it on assumption.

A positive birch-pollen or lipid-transfer-protein result is a different mechanism. Not every positive plant-food test is a storage-protein result. The same foods can also flag the milder birch-pollen-related proteins (which usually cause only an itchy mouth) or the lipid transfer proteins (a separate, sometimes cofactor-amplified pattern). Those are different proteins with a different prognosis, and they are covered on their own pages rather than here. The reason a positive storage-protein result is the one to take most seriously is precisely that it is not one of those milder patterns. Which protein lit up is the whole question, and that is what component testing answers.

These corrections clear specific fears (coconut, over-avoiding almond, mistaking a mild pollen result for a storage-protein one). They do not clear any food inside the very-high or high tiers above.

Where studies disagree

Two areas are genuinely unsettled, and the honest move is to publish the disagreement rather than pick a side. This is also what makes the “tests positive, usually tolerated” tier trustworthy instead of glib.

Co-sensitization versus real-world reactions. Across this whole cluster, a person allergic to one seed food very frequently tests positive to several others, because a broad panel co-fires across botanically related and even unrelated seeds. When those same people are challenged under supervision, real reactions to the distant members are much less common than the testing predicted. Both findings are true: the panel genuinely co-fires, and the clinical reactions genuinely do not follow at the same rate. The studies are measuring two different things, sensitization and reaction, and the gap between them is the whole reason the peanut-soy and chickpea-peanut links are “tests positive, usually tolerated” rather than “avoid everything that lights up.” The near-identical pairs are the exception that proves the rule: there the co-firing and the real reactions track closely, because the proteins are nearly the same molecule.

How far the legume links extend, especially lupin. The lentil-chickpea-pea triad clearly cross-reacts. The looser peanut-to-legume links (peanut-soy, chickpea-peanut) are mostly co-sensitization without reaction. Lupin sits in a contested middle: it is a documented real cross-reactor in legume-allergic people more often than a reassuring “low” framing would suggest, so this page treats lupin as a food to confirm with an allergist, not one to assume tolerance for. Where exactly the line falls for any individual legume is a question for testing and challenge, not a blanket rule.

Testing and confirmation: which protein lit up

Cross-reactivity questions in this cluster are answered by finding out which protein is driving the result.

A standard whole-extract blood test or skin prick tells you the immune system has noticed a seed food, but it does not tell you whether that is the dangerous storage-protein pattern or a milder one, and it does not separate a real allergy from a cross-reaction. Component-resolved testing breaks the result down protein by protein, so a result driven by a storage protein (the serious, cross-reactivity-meaningful kind) can be told apart from one driven by a birch-pollen-related or lipid-transfer protein. For the cross-reactivity question specifically, a positive result on the shared storage proteins is the signal that the very-high or high tier applies, and it is a red flag rather than a reassurance, as the mechanism section explains. One honest limit: most foods in this cluster do not have a single transferable decision number the way peanut’s Ara h 2 does, so your allergist reads the component result against your history rather than against a universal cutoff.

Where component testing and history still disagree, the supervised oral food challenge is the reference standard. It is the one test that turns “related on a panel” into “safe to eat” for a specific food, and it is done with your allergist, never at home, and never for a cross-reactive food on a hunch.

The members, each food’s own page:

The neighboring hubs:

Frequently asked questions

My child’s test came back positive for a storage protein. Is that a mild result?

No. A positive result on a seed storage protein (such as peanut Ara h 2, cashew Ana o 1 or Ana o 2, walnut Jug r 2 or Jug r 4, pistachio Pis v 2, Pis v 3 or Pis v 5, pecan Car i 2 or Car i 4, hazelnut Cor a 9 or Cor a 14) is a red flag for whole-body reactions, not a reassuring or minor finding, and these proteins are not destroyed by cooking. Treat it as a reason for strict avoidance and an epinephrine plan, and confirm any tolerance only with your allergist.

Why are cashew and pistachio, or walnut and pecan, treated as one allergy?

Because their storage proteins are near-identical. Cashew and pistachio share near-identical 2S albumins, and so do walnut and pecan, with roughly 9 in 10 people allergic to one of the walnut-pecan pair reacting to the other. When two seeds carry nearly the same protein, an allergy to one reliably predicts the other, so allergists manage each pair as a single avoidance.

I am allergic to peanut. Do I have to avoid soy and the other legumes?

Usually not, but it is tested, not assumed. Peanut and soy, and peanut and chickpea, commonly test co-positive, but clinical cross-reactivity is rare and most people with peanut allergy tolerate the other legumes; having one legume allergy does not mean avoiding all of them. Confirm with your allergist before introducing any of them. Lupin is the cautious exception: a peanut allergy is a reason to confirm lupin with your allergist rather than assume it.

Is coconut part of this cluster?

No. Coconut is botanically a fruit (a drupe), not a storage-protein seed, so it does not carry the proteins that drive this cluster, and most people with tree-nut allergy tolerate it. Confirm with your allergist before introducing it, but the tree-nut label overstates the botanical risk.

Does a positive test to one seed food mean I react to all of them?

No, not automatically. Within the cluster the strength tracks how alike the proteins are: the near-identical pairs (cashew/pistachio, walnut/pecan) reliably travel together, the related-protein axes (hazelnut with walnut and pecan, the lentil/chickpea/pea triad) are real and worth testing, and the distant links (peanut to soy or chickpea) usually test positive without a real reaction. A broad positive panel is a reason to test with your allergist, not a verdict that every seed food is dangerous.

Why is a storage-protein result more serious than a birch-pollen one?

Because they are different proteins. The seed storage proteins are heat-stable and survive digestion, so they tend to cause whole-body reactions, while the milder birch-pollen-related proteins are usually broken down by cooking and digestion and often cause only an itchy mouth. Which protein lit up is what decides the risk, which is why component testing is worth doing. The birch-pollen and lipid-transfer-protein patterns have their own pages.

References and medical review

This page is pending independent medical review until a named reviewer and date are in place. The verified cross-reactivity and reassurance claims resolve to the project’s conservative cross-reactivity floor, each carrying its own tier-1 source there. The component biology (the 2S albumin, 7S vicilin, and 11S legumin families, the heat- and digestion-stability mechanism, the co-sensitization-versus-reactivity gap, and the no-universal-cutoff finding for most cluster foods) resolves to the consolidated tree-nut, legume, and seed research reports still pending final review. The pecan-walnut “roughly 9 in 10” figure is published inside that verified floor record; the mustard storage-protein homology is given qualitatively as a near-identical 2S albumin, and figures not yet pinned to a stable source are omitted rather than stated. The almond-to-hazelnut, cashew-to-walnut, and pea-to-peanut edges are not established on the verified floor, so this page asserts them in neither direction.

← Cross-reactivity guides