12 Application of Computational Methods for the Safety Assessment …
251
to improve the predictivity of allergenicity from 3D structures [52], more research
is needed to identify and update the allergenic structural motifs and algorithms need
to be developed to predict structural similarity.
As mentioned previously, there is no specific guidance to determine the likelihood
of cross-reactivity based on full-length sequence homology. It has been proposed in
some recent publications [81, 82] to use the E-value in FASTA or PLASTP alignment, which represents the probability that the alignment might occur by chance,
to determine the degree of sequence homology. The developer of FASTA software
stated that an E-value of less than 10
−6 indicated a high certainty of homology [83].
However, because the E-value depends on the size of the database, there is currently
no clear regulatory guideline for using a threshold E-value in allergenicity assessment. Nevertheless, some scientists believe that using the E-value from full-length
sequence homology has a stronger scientific basis compared to the 80 AA sliding
window [53, 84]. Therefore, additional studies are needed to validate and standardize
the application of E-value in assessing protein allergenicity.
12.3.4 CERES Knowledgebase Expansion
A challenge in expanding the CERES knowledgebase is integrating data with those
from other sources that have different data structures (interoperability). Currently, the
chemical information in CERES, the OFAS’s submission repository system called
Food Application Regulatory management (FARM) system, and the CFSAN’s Scientific Terminology and Regulatory Information (STARI) system is being integrated
to create a centralized chemical information system in CFSAN, as well as to enable
efficient data abstractions from FARM and STARI into the CERES’s knowledgebase.
This data synchronization task is difficult as the three systems have different data
architectures. A temporary data model has been created to mitigate these challenges
and enforce data standardizations.
Another challenge in the knowledgebase expansion is data quality. At present, the
chemical, toxicity, and administrative data are required to go through external quality
control processes before being parsed into CERES. This mechanism can introduce
errors and misinterpretations as the data are manually transferred from one process
to another. Several data entry tools are being developed to address these issues. These
tools are intended to capture data directly from OFAS scientists within their standard
workflow process, thus minimizing data transfer errors and enforcing data standardizations. An example of such a tool is electronic memoranda (e-memos). E-memos are
Web forms that use the format of the official OFAS chemistry and toxicology memoranda that are written by the review scientists to summarize the technical reviews of
food ingredients. A typical chemistry review memorandum contains descriptions of
the chemical identity (chemical names and CASRN), intended use of the substance,
method of manufacture, and dietary intake values for the substance, along with the
chemistry reviewer’s comments/recommendations on the proposed use based on a
review of the chemistry information. A typical toxicology memorandum primarily
251
to improve the predictivity of allergenicity from 3D structures [52], more research
is needed to identify and update the allergenic structural motifs and algorithms need
to be developed to predict structural similarity.
As mentioned previously, there is no specific guidance to determine the likelihood
of cross-reactivity based on full-length sequence homology. It has been proposed in
some recent publications [81, 82] to use the E-value in FASTA or PLASTP alignment, which represents the probability that the alignment might occur by chance,
to determine the degree of sequence homology. The developer of FASTA software
stated that an E-value of less than 10
−6 indicated a high certainty of homology [83].
However, because the E-value depends on the size of the database, there is currently
no clear regulatory guideline for using a threshold E-value in allergenicity assessment. Nevertheless, some scientists believe that using the E-value from full-length
sequence homology has a stronger scientific basis compared to the 80 AA sliding
window [53, 84]. Therefore, additional studies are needed to validate and standardize
the application of E-value in assessing protein allergenicity.
12.3.4 CERES Knowledgebase Expansion
A challenge in expanding the CERES knowledgebase is integrating data with those
from other sources that have different data structures (interoperability). Currently, the
chemical information in CERES, the OFAS’s submission repository system called
Food Application Regulatory management (FARM) system, and the CFSAN’s Scientific Terminology and Regulatory Information (STARI) system is being integrated
to create a centralized chemical information system in CFSAN, as well as to enable
efficient data abstractions from FARM and STARI into the CERES’s knowledgebase.
This data synchronization task is difficult as the three systems have different data
architectures. A temporary data model has been created to mitigate these challenges
and enforce data standardizations.
Another challenge in the knowledgebase expansion is data quality. At present, the
chemical, toxicity, and administrative data are required to go through external quality
control processes before being parsed into CERES. This mechanism can introduce
errors and misinterpretations as the data are manually transferred from one process
to another. Several data entry tools are being developed to address these issues. These
tools are intended to capture data directly from OFAS scientists within their standard
workflow process, thus minimizing data transfer errors and enforcing data standardizations. An example of such a tool is electronic memoranda (e-memos). E-memos are
Web forms that use the format of the official OFAS chemistry and toxicology memoranda that are written by the review scientists to summarize the technical reviews of
food ingredients. A typical chemistry review memorandum contains descriptions of
the chemical identity (chemical names and CASRN), intended use of the substance,
method of manufacture, and dietary intake values for the substance, along with the
chemistry reviewer’s comments/recommendations on the proposed use based on a
review of the chemistry information. A typical toxicology memorandum primarily