12 Application of Computational Methods for the Safety Assessment …
243
tive information provided by the allergen databases used, which can vary by their
content, organization, and accessibility [49]. Currently, the most frequently used
database is the Food Allergy Research and Resource Program (FARRP) allergen
database (also known as the AllergenOnline, database, AOL), maintained at the University of Nebraska-Lincoln (http://www.allergenonline.org). Candidate proteins are
collected into the AOL database from the National Center for Biotechnology Information (NCBI) protein database, the World Health Organization/International Union
of Immunological Societies (WHO/IUIS) allergen nomenclature database, and peerreviewed publications on allergenicity. Each candidate protein is evaluated by an
expert panel in FARRP using a set of transparent predetermined criteria to determine whether it should be classified as an allergen and then added into the AOL
database [50]. The current version (Version 18B) of the AOL database includes 2089
AA sequences from 831 taxonomic-protein groups that are accepted with evidence
of allergic serum IgE-binding and/or biological activity. In addition to the AOL
database, there are several other online allergen databases available for bioinformatic usage, such as the UniProt (Previously referred to SwissProt) allergen database,
the Structural Database of Allergenic Proteins (SDAP), Allermatch, and Allfam (a
database of allergen families) [51, 52]. However, the peer-review process for each
of these databases has not been sufficiently defined and most of these databases are
not regularly updated, which limits their utility in bioinformatic analyses.
12.2.3.2 Criteria for Bioinformatic Analysis
Bioinformatic tools available online at AllergenOnline.org allow users to submit
the AA sequence of the query protein for overall or local sequence alignments,
searching for “matches” of identical short-sequence segments against the known and
putative allergenic proteins in the AOL database. Regarding the criteria of possible
IgE cross-reactivity, Codex recommends: (1) a threshold of 35% identity in a segment
(sliding window) of 80 or more AAs, corresponding to the typical size of a protein
domain containing IgE epitopes [53] and (2) identical sequence matches of six or
eight contiguous AAs of the query protein with an allergenic protein in the allergen
database. The FAO/WHO [54] recommends searching for a segment of six AAs,
considering six AAs as the minimal length of IgE recognizable sequence. However,
using only six AAs could return many false positives. Therefore, the AOL database
uses the 8-AA identity search criteria. Although Codex recommends full-length
sequence alignment using FASTA or BLASTP algorithms, a threshold degree of
homology that suggests allergenicity between the query protein and a known allergen
has not been clearly defined in the Codex guidelines. However, published analyses
have determined that at least 50–70% homology in proteins would be required to
show cross-reactivity [55].
While the scientific significance and sensitivity of the above-mentioned Codex
criteria have been challenged [53, 56–58], they are widely accepted and used for
allergenicity risk assessment by stakeholders. A negative sequence homology result
indicates that a protein is unlikely to be cross-reactive to a known allergen, whereas
243
tive information provided by the allergen databases used, which can vary by their
content, organization, and accessibility [49]. Currently, the most frequently used
database is the Food Allergy Research and Resource Program (FARRP) allergen
database (also known as the AllergenOnline, database, AOL), maintained at the University of Nebraska-Lincoln (http://www.allergenonline.org). Candidate proteins are
collected into the AOL database from the National Center for Biotechnology Information (NCBI) protein database, the World Health Organization/International Union
of Immunological Societies (WHO/IUIS) allergen nomenclature database, and peerreviewed publications on allergenicity. Each candidate protein is evaluated by an
expert panel in FARRP using a set of transparent predetermined criteria to determine whether it should be classified as an allergen and then added into the AOL
database [50]. The current version (Version 18B) of the AOL database includes 2089
AA sequences from 831 taxonomic-protein groups that are accepted with evidence
of allergic serum IgE-binding and/or biological activity. In addition to the AOL
database, there are several other online allergen databases available for bioinformatic usage, such as the UniProt (Previously referred to SwissProt) allergen database,
the Structural Database of Allergenic Proteins (SDAP), Allermatch, and Allfam (a
database of allergen families) [51, 52]. However, the peer-review process for each
of these databases has not been sufficiently defined and most of these databases are
not regularly updated, which limits their utility in bioinformatic analyses.
12.2.3.2 Criteria for Bioinformatic Analysis
Bioinformatic tools available online at AllergenOnline.org allow users to submit
the AA sequence of the query protein for overall or local sequence alignments,
searching for “matches” of identical short-sequence segments against the known and
putative allergenic proteins in the AOL database. Regarding the criteria of possible
IgE cross-reactivity, Codex recommends: (1) a threshold of 35% identity in a segment
(sliding window) of 80 or more AAs, corresponding to the typical size of a protein
domain containing IgE epitopes [53] and (2) identical sequence matches of six or
eight contiguous AAs of the query protein with an allergenic protein in the allergen
database. The FAO/WHO [54] recommends searching for a segment of six AAs,
considering six AAs as the minimal length of IgE recognizable sequence. However,
using only six AAs could return many false positives. Therefore, the AOL database
uses the 8-AA identity search criteria. Although Codex recommends full-length
sequence alignment using FASTA or BLASTP algorithms, a threshold degree of
homology that suggests allergenicity between the query protein and a known allergen
has not been clearly defined in the Codex guidelines. However, published analyses
have determined that at least 50–70% homology in proteins would be required to
show cross-reactivity [55].
While the scientific significance and sensitivity of the above-mentioned Codex
criteria have been challenged [53, 56–58], they are widely accepted and used for
allergenicity risk assessment by stakeholders. A negative sequence homology result
indicates that a protein is unlikely to be cross-reactive to a known allergen, whereas