Extending OpenTox to Nano Safety with eNanoMapper
Christoph Helma
in silico toxicology gmbh
Basic idea
Webservices as building blocks for computational toxicology
- avoid reinventing the wheel
- reuse existing components
- rapid prototyping
- interoperability
- uses semantic web technologies
Application programming interface (API)
- Request messages:
- Representational state transfer (REST, based on HTTP messages)
- Data exchange:
- Structure: object dependent
- Format: Resource Description Framework (RDF), JavaScript Object Notation (JSON)
OpenTox
Focus on predictive toxicology
Main objects: Compound, Feature, Dataset, Algorithm, Model, Validation
Helper objects: Task, Authentication and Authorisation, Report, Ontology
Data format: RDF
ToxBank
Extensions for capturing bioassay data
Main addition: Investigation (description of biological experiments)
Helper objects: Alert, Data, Index, Organisation, Project, Protocol, Search, Session, Template, User
Data format: RDF, ISA-Tab
eNanoMapper
Extensions for capturing nanomaterial data
Main addition: Substance (characterisation of nanoparticles)
Helper objects: Bundle, Myaccount, Property, Query, Substanceowner, Pmml, Bibtex
Data format: JSON
eNanoMapper APIs
IDEA: http://enanomapper.github.io/API
Technical problems
- (De)Serialisation of large datasets/investigations
- Reliability/availability/compatibility of external services
- Local service deployment (e.g. for in house services)
Documentation problems
- Scattered/outdated documentation
- Difficult to maintain for developers (no suitable cross-platform solution for code-based documentation)
- Limitations of documentation platforms (e.g. SWAGGER) to document important details
API problems
- Lack of orthogonality (large number of partially redundant helper objects)
- Ad hoc extensions of individual developers
- Undocumented/silent API/data format changes
Coordination problems
- Individual API changes may be necessary and justified
- Conflict resolution works one to one, but leaves out other developers (and consistency)
- No common strategy for conflict resolution and API updates
- Very limited developer base for testing and feedback