6 Mode-of-Action-Guided, Molecular …
107
the modification of S target by modulators, and S adme accounts for the influence of
ADME on chemical bioavailability biomacromolecules.
6.3 Results
As an ongoing project, we divide the development of our novel MoA/MIE-guided,
molecular modeling-based approach for in silico predictive toxicology into two major
phases. In Phase 1, we intend to focus on prediction of in vitro toxicity endpoints.
In Phase 2, we will expand to in vivo toxicity end points by incorporating ADME
processes into prediction modeling (Fig. 6.1). In the following, we provide an update
on the Phase 1 status of the TsTKb development. For more details about the TsTKb,
one may refer to our recent publication [50].
6.3.1 Libraries of Reference Chemicals in ChemMoA
We have curated a library of reference chemicals for each MoA/MIE or toxicity
target according to the following criteria: (1) the availability of toxicity data, (2)
the uniqueness of toxicity target (to avoid chemicals interacting with multiple targets within a toxicity pathway and eliciting the same toxicity at the organ/system
level), and (3) their reported toxicity spanning a wide potency spectrum. We have
queried more than a dozen publicly accessible databases: Aggregated Computational
Toxicology Online Resource (ACToR) [54] and Distributed Structure-Searchable
Toxicity (DSSTox) [55] databases, both developed by the US EPA; PubChem [56];
ChEMBL [57]; ZINC15 [58, 59]; Estrogenic Activity Database (EADB), Endocrine
Disruptor Knowledge Base (EDKB), and Liver Toxicity Knowledge Base (LTKB),
all of which were developed by researchers at the US FDA [60]; SuperTarget [61];
SuperToxic [62]; Toxin and Toxin Target Database (T3DB) [63, 64]; TG-GATE
[65]; and TOXNET [66, 67]. We retain chemicals that cause a wide variety of toxic
effects (e.g., acetolactate synthase inhibition, GABA A receptor antagonism, hepatic
steatosis, acetylcholinesterase inhibition, androgen receptor antagonism/agonism,
and estrogen receptor antagonism/agonism) through interacting with their respective
toxicity targets. MoA/MIE data for these chemicals are retrieved and categorized,
whereas toxicity data are normalized and harmonized. The AOP knowledgebase,
developed as part of the OECD AOP Development Effort [18, 68] and hosted at the
AOP Wiki Web portal (https://aopwiki.org/) is also consulted with regard to MoAbased chemical categorization. An example of the curated hepatotoxin library is
provided in Fig. 6.2.
107
the modification of S target by modulators, and S adme accounts for the influence of
ADME on chemical bioavailability biomacromolecules.
6.3 Results
As an ongoing project, we divide the development of our novel MoA/MIE-guided,
molecular modeling-based approach for in silico predictive toxicology into two major
phases. In Phase 1, we intend to focus on prediction of in vitro toxicity endpoints.
In Phase 2, we will expand to in vivo toxicity end points by incorporating ADME
processes into prediction modeling (Fig. 6.1). In the following, we provide an update
on the Phase 1 status of the TsTKb development. For more details about the TsTKb,
one may refer to our recent publication [50].
6.3.1 Libraries of Reference Chemicals in ChemMoA
We have curated a library of reference chemicals for each MoA/MIE or toxicity
target according to the following criteria: (1) the availability of toxicity data, (2)
the uniqueness of toxicity target (to avoid chemicals interacting with multiple targets within a toxicity pathway and eliciting the same toxicity at the organ/system
level), and (3) their reported toxicity spanning a wide potency spectrum. We have
queried more than a dozen publicly accessible databases: Aggregated Computational
Toxicology Online Resource (ACToR) [54] and Distributed Structure-Searchable
Toxicity (DSSTox) [55] databases, both developed by the US EPA; PubChem [56];
ChEMBL [57]; ZINC15 [58, 59]; Estrogenic Activity Database (EADB), Endocrine
Disruptor Knowledge Base (EDKB), and Liver Toxicity Knowledge Base (LTKB),
all of which were developed by researchers at the US FDA [60]; SuperTarget [61];
SuperToxic [62]; Toxin and Toxin Target Database (T3DB) [63, 64]; TG-GATE
[65]; and TOXNET [66, 67]. We retain chemicals that cause a wide variety of toxic
effects (e.g., acetolactate synthase inhibition, GABA A receptor antagonism, hepatic
steatosis, acetylcholinesterase inhibition, androgen receptor antagonism/agonism,
and estrogen receptor antagonism/agonism) through interacting with their respective
toxicity targets. MoA/MIE data for these chemicals are retrieved and categorized,
whereas toxicity data are normalized and harmonized. The AOP knowledgebase,
developed as part of the OECD AOP Development Effort [18, 68] and hosted at the
AOP Wiki Web portal (https://aopwiki.org/) is also consulted with regard to MoAbased chemical categorization. An example of the curated hepatotoxin library is
provided in Fig. 6.2.