|
Microsoft Research selected 6 proposals for the
"Computational Tools for Advancing Science" call for proposals. We
were particularly
interested in new tools that advance knowledge discovery, facilitate knowledge
sharing, and enable decision support, hypothesis and model generation and
testing, visualisation and the ability to automatically, intelligently perform,
interpret and draw conclusions from complex experiments.
Award Recipients
Adding new dimensions to SkyServer
István Csabai, Roland Eötvös University, Hungary
Based on the new technologies that enable the integration of C like code
with databases a high dimensional spatial indexing and data mining tools
will be developed. The scientific dataset on which this will be implemented
and on which the methods will be demonstrated is the SDSS SkyServer.
Computational platform for modelling of signal transduction
Roland Eils, German Cancer Research Centre
(DKFZ), Germany
Hans Georg Bock, University of Heidelberg, Interdisciplinary
Centre for Scientific Computing, Germany
The aim of this project is to develop computational support for system
identification in signal transduction. In particular, tools will be developed
to estimate the unknown system parameters in non-linear dynamic models from
microscopic images in live-cell experiments. Moreover, methods for the optimal
design of such experiments will be elaborated in order to increase the information
content of the system maximally while minimising the experimental costs
involved. The tools will be implemented and integrated into a software platform
that supports the identification and analysis of spatial dynamic models
of large-scale signalling networks.
Geometric intelligence
Bruno Lévy, INRIA Lorraine, France
The main objective of this project is to design new solutions to convert
a raw representation of a 3D object into a higher-level representation.
More specifically, this means designing algorithms that
"understand the
geometry. To recover the high level structures of the objects, the approach
will consist in writing a complete mathematical formalisation of the problem
and in restating the geometric problem in terms of numerical optimisation.
The approach will bridge the gap between several previously incompatible
representations at a fundamental level.
Predictive modelling of signalling pathways via probabilistic model
checking with PRISM
Marta Kwiatkowska, University of Birmingham, United Kingdom
This project will contribute to the predictive biology effort by adapting
process calculi and the probabilistic model checking technology, and specifically
the leading probabilistic model checker PRISM developed at the University
of Birmingham, to the study of biological processes. This complements the
traditional differential equations (ODE) modelling approaches and offers
new ways to affirm or disprove scientific hypotheses.
Self-organising data stream management for e-science
Alfons Kemper, Technical University of Munich, Germany
Recent research efforts in the fields of data stream processing and data
stream management systems (DSMS) show the increasing importance of processing
data streams, e.g., in the context of e-science applications. The project
will develop novel approaches for handling and processing data streams,
e.g., for incrementally registering new continuous queries over data streams
in P2P networks, thus optimising data flow, reducing network traffic and
avoiding redundant computation of stream transforming operators. The goal
is to support efficient on-the-fly dissemination, evaluation and (possibly)
reduction of experimental or observational data in the e-science domain.
The research will be conducted in close cooperation with astrophysical scientists.
Towards automated conceptualisation and accelerated knowledge discovery
in molecular dynamics
Christof Schuette, Freie Universität Berlin, Germany
Hans-Christian Hege, Konrad-Zuse-Institute Berlin (ZIB),
Germany
This project intends to start the development of a first software platform
based on a generalised database of molecular dynamics (MD) time series and
offering advanced interactive techniques for statistical and time series
analysis. The platform will allow the users to visually illustrate reduced
dynamical information and it will offer tools for conceptualising MD results
into physical knowledge about non-equilibrium properties. The intended research
is based on experience of the partners in non-equilibrium modelling of (bio)molecular
systems, time series analysis, statistical learning, feature extraction,
data visualisation, as well as construction, implementation, and distribution
of integrated software platforms.
|
