COSMOS DataShare Point: Update and Focus on … · COSMOS DataShare Point: Update and Focus on Chemical Inventories SOT 2017 Baltimore Mark Cronin Liverpool John Moores University,

COSMOS DataShare Point: Update and Focus on Chemical

Inventories

SOT 2017 Baltimore

Mark Cronin

Liverpool John Moores University, England

In Silico Models

Project (2011-2015)

Threshold of Toxicological Concern (TTC)

PBPK and In Vitro – In Vivo Extrapolation

New Toxicological Databases

COSMOS DB: Freely Available and Searchable Resources for Toxicity Data

• Open-source technology

• High quality data

• Intuitive browsing of complex data

COSMOS DB v2.0

• Publicly available

• Houses data for human health endpoints

• Building blocks for safety evaluation

• Provides data relevant to international regulation of cosmetics-related chemicals

• Maintained through COSMOS DataShare Point initiative

Content of COSMOS DB v2.0

• 81,602 COSMOS IDs

• 44,765 unique structures

• Cosmetics Inventory • INCI names

• Chemical functions

• 12,538 toxicity studies for 1,660 compounds across 27 endpoints

• oRepeatTox DB toxicity data (230 cosmetics-related chemicals and 340 studies)

COSMOS DB v2.0: Safety Evaluation and TTC Database, Decision Tree

DataShare Point

https://cosmosdb.eu/cosmosdb.v2

• An opportunity to share data

– Focus on mammalian ADMET endpoints

– Rich legacy of repeat-dose data

– Chemical inventories

• High quality, curated structures and toxicity data

• Detailed study information captured

• Informed by toxicologists

• Maintain, update and extend current data and functionalities of COSMOS DB v2

• A publicly available collaborative resource

• Linkage to chemoinformatics technologies

DataShare Point

COSMOS DataShare Point:

Why Get Involved • It’s not just another data sharing exercise

• Free access to data

– Building unique databases

• Supported by high quality chemical structures

– Cosmetics inventory

– Essential to support modelling and data capture

• Allows for data mining

– Organ level effects

• Data quality addressed

– Data entry system

Contributions

• National Institute of Technology and Evaluation (Japan)

– HESS DB – approx. 500 GLP repeat dose toxicity studies

• NCTR/ US FDA (Weida Tong)

– Liver (3000); Carcinogenicity (1000); ED (8200)

• EC JRC

– ECVAM HTS, liver metabolism, additional datasets

• US Cosmetics Ingredient Review (CIR)

– Safety reports 4,600 ingredients

• Korean Cosmetic Industry Institute (KCII)

– Safety evaluation database

• eTOX IMI Project

– Preclinical repeat dose toxicity data

Chemical Inventories: GPS

• Many toxicity data associated with these inventories (although not necessarily in COSMOS DB)

Focus on Repeat Dose Toxicity Data in

COSMOS DB: Toxicity Data Sources

number of studies

(compounds)

US FDA CFSAN OFAS 98

US FDA CFSAN PAFA 68 (45)

SCCNFP/SCCP/SCCS 126 (112)

ECHA (Registered Substances Database) 68 (30)

US EPA (ToxRefDB, Data Evaluation Record) 65 (39)

US EPA (IRIS) 7 ( 6)

US National Toxicology Program 9 (9)

EFSA (European Food Safety Agency) 3 (3)

Literature 8 (8)

• US FDA PAFA legacy database

• Food direct and indirect additives, colorants

• oRepeatToxDB

Liver Toxicity – Cosmetics vs Tox21

compounds in oral

repeat dose DB

also in COSMOS

DB with liver

effects

plus other

compounds in oral

repeat dose DB also

in COSMOS DB

plus

remaining

compounds in

COSMOS DB

Properties: molecular weight, # H donors, # H acceptors, XlogP, TPSA, polarizability, dipole moment, logS, rule-of-

five violations, complexity, ring complexity, diameter

Principal Component Projections Based on Properties:

Sample of COSMOS DB / Oral Repeat Dose

COSMOS (3369)

Tox21 (6394)

Principal Component Projections Based on Properties:

COSMOS DB vs Tox21

COSMOS (3369)

Tox21 (6394)

Principal Component Projections Based on ToxPrint

Chemotypes: COSMOS DB vs Tox21

-5

0

5

-4-2024-3

-2

-1

0

1

2

3

PC

4

PC3

PC2

COSMOS Inventory

Pharmaceuticals

Pharmaceuticals and Cosmetics

Separation by ToxPrints

A core for small generic structural features appearing in all organic chemicals. Cannot be considered as overlapping chemical space

COSMOS inventory set: 4459 structure | pharmaceuticals: 908 structures

Properties: molecular weight, # H donors, # H acceptors, XlogP, TPSA,

polarizability, dipole moment, logS, rule-of-five violations, complexity, ring

complexity, diameter

-5

0

5

10

15

-6-4-20246

-5

0

5

PC

4

PC3

PC2

Pharmaceuticals and Cosmetics:

Separation by Physicochemical Properties

• There are some extreme

values of properties in

cosmetics.

• The drugs cluster only

shows the loading on the

positive scores of PC3

whilst the cosmetics

appear both as positive

and negative scores in

PC3.

New for 2017

• Cancer Potency Database (CPDB)

– Extended through CEFIC LRi Project

• Chemical inventory creation

– Exploring “chemical space”

– GPS Concept

– Mapping effects onto “chemical space”

Future Plans

• Continue to develop COSMOS DB v2 and its functionalities

• Enter data through collaborative agreements

• Mine and utilise data in new ways

• Maintain inventories of chemical IDs

• Developing better knowledge

• Analysis of the data, located in specific regions of chemical space is now possible

• Updates through COSMOS DataShare Point Fora – SOT and Eurotox

Thank You and Any Questions

DataShare Point

https://cosmosdb.eu/cosmosdb.v2

Acknowledgements

• The European Community’s Seventh Framework Program (FP7/2007-2013) COSMOS Project under grant agreement n° 266835 and Cosmetics Europe

• Co-workers in Liverpool, EU, USA