Advisory self-classifications for 54,135 substances based on (Q)SAR predictions from the Danish (Q)SAR database, VEGA QSAR and the OECD QSAR Toolbox

Nikolai Georgiev Nikolov, Henrik Tyle, Magnus Løfstedt, Eva Bay Wedebye

Research output: Contribution to conferenceConference abstract for conferenceResearchpeer-review

29 Downloads (Pure)

Abstract

Lack of experimental data on toxicological properties make it difficult for companies to self-classify the chemical substances they import or produce. To address this issue, 80,085 pre-registered and/or registered REACH substances from the Danish (Q)SAR Database were evaluated by (Q)SAR. For this purpose predictions primarily from the Danish (Q)SAR Database were used, supplemented with predictions from VEGA QSAR and a number of the profilers available from the OECD QSAR Application Toolbox. The following classification endpoints were addressed:

• Mutagenicity: Muta. 2
• Carcinogenicity: Carc. 2
• Reproductive toxicity (possible harm to the unborn child): Repr. 2
• Acute oral toxicity: Acute Tox.1-4
• Skin irritation: Skin Irrit. 2
• Skin sensitisation: Skin Sens. 1
• Danger to the aquatic environment: Acute 1, Chronic 1-3

Algorithms were developed for each classification endpoints to combine predictions to reach a final call in an attempt to reach further reliability and to best comply with the classification criteria. No advisory predictions were based on a positive prediction from a single system, and if only based on a battery prediction (majority vote from 3 systems) the third system was required not to give a negative prediction in applicability domain. This resulted in a list with a total of 54,135 substances with one or more advisory self-classifications. The list is available from the Danish Environmental Protection Agency homepage.
Original languageEnglish
Publication date2018
Number of pages2
Publication statusPublished - 2018
EventQSAR2018: 18th International Conference on QSAR in Environmental and Health Sciences - Rikli balance hotel , Bled, Slovenia
Duration: 11 Jun 201815 Jun 2018
Conference number: 18th
http://www.qsar2018.com/

Conference

ConferenceQSAR2018
Number18th
LocationRikli balance hotel
CountrySlovenia
CityBled
Period11/06/201815/06/2018
Internet address

Cite this

@conference{8739f7c61b3a446f984ac097665d07c4,
title = "Advisory self-classifications for 54,135 substances based on (Q)SAR predictions from the Danish (Q)SAR database, VEGA QSAR and the OECD QSAR Toolbox",
abstract = "Lack of experimental data on toxicological properties make it difficult for companies to self-classify the chemical substances they import or produce. To address this issue, 80,085 pre-registered and/or registered REACH substances from the Danish (Q)SAR Database were evaluated by (Q)SAR. For this purpose predictions primarily from the Danish (Q)SAR Database were used, supplemented with predictions from VEGA QSAR and a number of the profilers available from the OECD QSAR Application Toolbox. The following classification endpoints were addressed:• Mutagenicity: Muta. 2 • Carcinogenicity: Carc. 2 • Reproductive toxicity (possible harm to the unborn child): Repr. 2• Acute oral toxicity: Acute Tox.1-4 • Skin irritation: Skin Irrit. 2• Skin sensitisation: Skin Sens. 1 • Danger to the aquatic environment: Acute 1, Chronic 1-3Algorithms were developed for each classification endpoints to combine predictions to reach a final call in an attempt to reach further reliability and to best comply with the classification criteria. No advisory predictions were based on a positive prediction from a single system, and if only based on a battery prediction (majority vote from 3 systems) the third system was required not to give a negative prediction in applicability domain. This resulted in a list with a total of 54,135 substances with one or more advisory self-classifications. The list is available from the Danish Environmental Protection Agency homepage.",
author = "Nikolov, {Nikolai Georgiev} and Henrik Tyle and Magnus L{\o}fstedt and Wedebye, {Eva Bay}",
year = "2018",
language = "English",
note = "QSAR2018 : 18th International Conference on QSAR in Environmental and Health Sciences, QSAR2018 ; Conference date: 11-06-2018 Through 15-06-2018",
url = "http://www.qsar2018.com/",

}

Advisory self-classifications for 54,135 substances based on (Q)SAR predictions from the Danish (Q)SAR database, VEGA QSAR and the OECD QSAR Toolbox. / Nikolov, Nikolai Georgiev; Tyle, Henrik; Løfstedt, Magnus; Wedebye, Eva Bay.

2018. Abstract from QSAR2018, Bled, Slovenia.

Research output: Contribution to conferenceConference abstract for conferenceResearchpeer-review

TY - ABST

T1 - Advisory self-classifications for 54,135 substances based on (Q)SAR predictions from the Danish (Q)SAR database, VEGA QSAR and the OECD QSAR Toolbox

AU - Nikolov, Nikolai Georgiev

AU - Tyle, Henrik

AU - Løfstedt, Magnus

AU - Wedebye, Eva Bay

PY - 2018

Y1 - 2018

N2 - Lack of experimental data on toxicological properties make it difficult for companies to self-classify the chemical substances they import or produce. To address this issue, 80,085 pre-registered and/or registered REACH substances from the Danish (Q)SAR Database were evaluated by (Q)SAR. For this purpose predictions primarily from the Danish (Q)SAR Database were used, supplemented with predictions from VEGA QSAR and a number of the profilers available from the OECD QSAR Application Toolbox. The following classification endpoints were addressed:• Mutagenicity: Muta. 2 • Carcinogenicity: Carc. 2 • Reproductive toxicity (possible harm to the unborn child): Repr. 2• Acute oral toxicity: Acute Tox.1-4 • Skin irritation: Skin Irrit. 2• Skin sensitisation: Skin Sens. 1 • Danger to the aquatic environment: Acute 1, Chronic 1-3Algorithms were developed for each classification endpoints to combine predictions to reach a final call in an attempt to reach further reliability and to best comply with the classification criteria. No advisory predictions were based on a positive prediction from a single system, and if only based on a battery prediction (majority vote from 3 systems) the third system was required not to give a negative prediction in applicability domain. This resulted in a list with a total of 54,135 substances with one or more advisory self-classifications. The list is available from the Danish Environmental Protection Agency homepage.

AB - Lack of experimental data on toxicological properties make it difficult for companies to self-classify the chemical substances they import or produce. To address this issue, 80,085 pre-registered and/or registered REACH substances from the Danish (Q)SAR Database were evaluated by (Q)SAR. For this purpose predictions primarily from the Danish (Q)SAR Database were used, supplemented with predictions from VEGA QSAR and a number of the profilers available from the OECD QSAR Application Toolbox. The following classification endpoints were addressed:• Mutagenicity: Muta. 2 • Carcinogenicity: Carc. 2 • Reproductive toxicity (possible harm to the unborn child): Repr. 2• Acute oral toxicity: Acute Tox.1-4 • Skin irritation: Skin Irrit. 2• Skin sensitisation: Skin Sens. 1 • Danger to the aquatic environment: Acute 1, Chronic 1-3Algorithms were developed for each classification endpoints to combine predictions to reach a final call in an attempt to reach further reliability and to best comply with the classification criteria. No advisory predictions were based on a positive prediction from a single system, and if only based on a battery prediction (majority vote from 3 systems) the third system was required not to give a negative prediction in applicability domain. This resulted in a list with a total of 54,135 substances with one or more advisory self-classifications. The list is available from the Danish Environmental Protection Agency homepage.

M3 - Conference abstract for conference

ER -