You are here:
Atomic contribution mapping and exploration with reverse fingerprinting (ACME-RF): Assigning toxicological endpoints to chemical structure at atomic resolution
Citation:
Goldsmith, M. AND C. Williams. Atomic contribution mapping and exploration with reverse fingerprinting (ACME-RF): Assigning toxicological endpoints to chemical structure at atomic resolution. QSAR 2021 International Workshop on QSAR in Environmental and Health Sciences, Virtual, NC, June 07 - 10, 2021.
Impact/Purpose:
Abstract submitted to the QSAR 2021 International Workshop on QSAR in Environmental and Health Sciences June 2021. In silico tools and models for assessing activity are usually defined by endpoints and quantitative structural metrics. Although it is useful to obtain categorical/continuous estimates of activity, traditional SAR provide limited guidance as to the molecular moieties giving rise to the endpoint. Reverse-fingerprinting (RF), provides a useful marriage between discretized endpoints and feature-based molecular fingerprint. RF produces both a quantitative and visual representation of atomic contribution to an endpoint, mapped on to structure (Williams C, 2009 PMID: 19442069). Here we introduce the concept of atomic contribution mapping and exploration (ACME) using the RF framework.
Description:
In silico tools and models for assessing activity are usually defined by endpoints and quantitative structural metrics. Although it is useful to obtain categorical/continuous estimates of activity, traditional SAR provide limited guidance as to the molecular moieties giving rise to the endpoint. Reverse-fingerprinting (RF), provides a useful marriage between discretized endpoints and feature-based molecular fingerprint. RF produces both a quantitative and visual representation of atomic contribution to an endpoint, mapped on to structure (Williams C, 2009 PMID: 19442069). Here we introduce the concept of atomic contribution mapping and exploration (ACME) using the RF framework. Using public datasets, we explore three different ACME-RF examples. First, we demonstrate the rapid identification of a class of pyrethroid acaricide that is not-toxic to honeybees while still being toxic to the varroa mite using very basic insecticide-class information of 80 pyrethoids as inputs. Second, we used the ToxCast NVS_NR_hER dataset (165/2645) to build a RF model that was used to identify the toxicophore of hER-a that directly map to known crystal structures. Finally, we explore photostability half-lives (Blum, Kristin M. 2013) and identify critical photolabile moieties. Using ACME-RF we identified and visualized moieties of molecules that resulted in (I) apical endpoints across species (II) chemical-biological interactions and (III) photodegradation liabilities. The method can be used to identify toxic chemicals and critical toxicophore fragments or sub-structures essential for molecular discovery and de-risking. This abstract does not necessarily reflect the views or policy of the US EPA.