BIOCLIPSE-OPENTOX:
INTERACTIVE PREDICTIVE TOXICOLOGY
Nina Jeliazkova a, Ola Spjuth b, Egon Willighagen c
a) Ideaconsult Ltd., Sofia, Bulgaria, jeliazkova.nina@gmail.com, twitter: @10705013
b) Department of Pharmaceutical Bioinformatics, Uppsala University, Uppsala, Sweden
c) Department of Bioinformatics - BiGCaT, Maastricht University, The Netherlands, egon.willighagen@maastrichtuniversity.nl, twitter: @egonwillighagen
OpenTox Computational Toxicologist wish list Bioclipse
• A semantic web framework, based on web 1. Uniform access to the (online) databases with • A desktop bio- and cheminformatics
services [1,2] http://opentox.net experimental data and to computational platform [3]
methods
• Supports open data exchange and • http://bioclipse.net/
2. Interactively, graphically explore chemicals
toxicology model building and validation • Combines a rich scriptable and graphical
3. See what parts (atoms, fragments) of the
Uniform access via molecule trigger toxic responses workbench environment
OpenTox web services API 4. Integrate with data analysis methods
Visualization
5. Allow reproducibility of the analysis
to Chemical structures and Datasets
• The representation of data and processing Result: Bioclipse – OpenTox [1-4]
resources in W3C Resource Description
Framework facilitates integrating the OpenTox web services can be used as:
resources as Linked Data.
• components in web applications
• By uploading datasets with chemical
structures and arbitrary set of properties, • components in desktop applications
they become automatically available • embedded in existing workflow systems
online in several formats.
to Computational Components
• descriptor calculations
• machine learning algorithms The toxicity is often determined by substructures
• similarity/structure searching
algorithms
Scripting, workflows,
• applicability domain
interactivity
• toxicity prediction models
for building and publishing Despite the workflows flexibility, some
predictive models users prefer simple user interfaces or
software with predefined functionality,
1. Start with an OpenTox dataset:
while some experts consider scripting
Computational methods discovery
languages more convenient for model
• Bioclipse dynamically discovers computational development than graphical workflow
algorithms exposed via the OpenTox servers [4], builders.
using the OpenTox ontology service's SPARQL Bioclipse provides convenient solution for
endpoint. both groups of users, offering
• The SPARQL endpoint functions as a registry of simultaneously GUI actions, visualizations
available data and computational services on and scripts.
the OpenTox network, similar to the role of
BioCatalogue.
• When a new descriptor algorithm or model is
registered on the OpenTox ontology service, it
2. Launch a machine learning algorithm via
Bioclipse JavaScript: will automatically be picked up by Bioclipse.
• Using this approach, Bioclipse has access to the
most recent computational services running the
latest predictive algorithms and models, while
hiding technicalities by reusing a graphics-
oriented workbench for the life sciences.
3. The result is an online regression or
classification model, available via OpenTox web
Bioclipse-OpenTox interoperability for
service API: Computational toxicology can be
online descriptor calculation and data
made reproducible using an application
sharing , as well as accessing protected
programming interface (API) for models
OpenTox datasets and calculation
and data and scripting.
procedures via the adopted OpenSSO
authentication and authorization The scripting functionality makes it easy to
solution had been demonstrated in [4]. automate data workflows as do workflow
Bioclipse scripting was recently extended applications but the combination with the
with the ability to build new OpenTox rich Bioclipse user interface makes it
models and thus to add new online possible at the same time to work with
computational resources. OpenTox interactively.
References
1) Hardy B., Douglas N., Helma C., Rautenberg M., Jeliazkova N., et al. , Collaborative Development of Predictive Toxicology Applications, Journal of Cheminformatics 2010, 2:7
2) Jeliazkova N., Jeliazkov V. AMBIT RESTful web services: an implementation of the OpenTox application programming interface., Journal of Cheminformatics 2011, 3:18.
3) Spjuth, O., Helmus, T., Willighagen, E. L., Kuhn, S., Eklund, M., Wagener, J., Murray-Rust, P., Steinbeck, C., and Wikberg, J. E. S. Bioclipse: an open source workbench for chemo- and bioinformatics.
BMC Bioinformatics 2007, 8:59.
4) Willighagen, E.L., Jeliazkova, N., Hardy, B., Grafström, R.C., Spjuth, O. Computational toxicology using the OpenTox application programming interface and Bioclipse, BMC Research Notes 2011,
4:487.