Figure 2. Screenshots of the web interface when used for retrieving the symptoms associated to the major birch pollen allergen (Bet v 1 ). The left panel shows the query form where the user can enter the allergen or symptoms to search for (in this case the name of the allergen) and the type of relationships. The right panel shows the results of the query, in this case the list of symptoms for Bet v 1 .
To identify those allergens potentially causing life-threatening, we selected a series of severe symptoms which are disclosed in decreasing order of frequency:
  1. anaphylactic shock (HP:0100845) or anaphylaxis (D000707) associated to 210 individual allergens;
  2. exercise induced anaphylaxis (HP:0410139) associated to 34 individual allergens;
  3. food-induced anaphylaxis (HP:0500095) associated to 27 individual allergens;
  4. dyspnea (HP:0002094) associated to 10 individual allergens;
  5. idiopathic anaphylaxis (HP:0410148) associated to 6 individual allergens;
  6. loss of consciousness (HP:0007185) associated to 5 individual allergens;
  7. hypotension (HP:0002615) associated to 4 individual allergens;
  8. Collapse associated to 3 individual allergens;
Overall, 222 allergens could be significantly correlated with life-threatening severe allergic reaction symptoms (Supplementary Table 1), whereas 814 allergens were statistically correlated to a wide range of less severe symptoms.
Regarding the main exposure route of those allergens correlated with life-threatening severe allergic reaction symptoms, 121 allergens were related to ingestion, 51 to insect venom allergies, 44 were classified as aeroallergens (23 of them being derived from plants, 10 to cats, dogs, horses or guinea pigs, 9 derived from dust mites, 2 corresponding to american cockroach) and 6 related to dermal contact (mostly to latex allergens) (Supplementary Table 1).
This list of 222 allergens was further manually refined based on the following prioritization criteria:
i) consideration of ingestion, contact and airway exposure routes of allergens related to food sources. This criterium served to disregard, mainly, insect venom allergens, and some aeroallergens from sources without any food relevance;
ii) consideration of allergens belonging to sources recognised as the most common causing food allergies and intolerances based on European Regulation[21] and the reviewed Codex priority allergen list by an ad hoc Joint FAO/WHO Expert Consultation Committee[22].
iii) consideration of allergens belonging to well-known allergen protein families or subfamilies (e.g ., tropomyosins, parvalbumins, non-specific lipid-transfer proteins, 2S albumins, vicilins, legumins) regardless of the food relevance of the source; or
iv) consideration of allergens not included in any of the above criteria but unequivocally associated to life-threatening severe allergic symptoms. This last criterium served to disregard a few allergens included in the initial list as they could not be directly linked to severe symptoms. This was due to several reasons, such as:
- some of the significant co-mentions retrieved in certain articles were based on the correlation of anaphylaxis with the whole food source instead of specific allergens,
- some articles compared allergens associated with a wide range of symptoms and the text-mining approach could not automatically differentiate among those allergens associated to severe symptoms and those related to mild symptoms,
- some articles dealt with allergens and/or derived peptides used for immunotherapy but these proteins were not identified as responsible for triggering severe symptoms.
Overall, a refined list containing 137 food allergens resulted from applying above criteria (Table 1 ). This set of allergens could be further considered for a more target allergenicity RA of novel proteins as will be discussed below.