You are here: Home Collaborations Developer Notes COSMOS

COSMOS Initial meeting - Minutes

Minute of collaboration initial meeting. 2011

Course of meeting

Present were Ulrich Sternberg, Olaf Schneider, and Ivan Kondov from Karlsruhe Institute of Technology, and Rasmus Fogh, Wayne Boucher, Tim Stevens and John Ionides fro the University of Cambridge. The two sides presented their respective programs (COSMOS and the CCPN / CcpNmr suite), and discussed the objectives and modalities of integration. The following was agreed:


Strategic objectives are:
COSMOS: Reaching new communities of users and improving visibility and accessibility of COSMOS.
CCPN: Adding COSMOS to the set of software accessible from CCPN data model, and using its capabilities to improve functionality of CcpNmr suite.

COSMOS should be integrated with the standard WmsProtocol and NmrCalc data storage, and the standard interface derived from it (CCPN). Data transfer shall be in memory. As a first step, the integration should use a local installation. The integration should be done at a meeting in Cambridge in July (to be arranged), with as much work as possible done beforehand, as drafts. In the longer term there will be integration thbrough the web-based Wms workflow manager, and a COSMOS web service, callable from either integration.

Integration should hopefully be done with Olaf Schneider, but he is moving to a different department and must wait a bit to arrange meeting. If he cannot make it, Ulrich Steinberg will step in.

At first integration should focus on doing geometry optimisation and calculating chemical shifts for a given protein structure. This can be integrated in Analysis for validation, highlighting of problematic parts of the protein, assignment aid, and selecting structures from an ensemble.


  • Integration will be in memory. It is planned that it will be done by Karlsruhe Data will be transferred as in-memory representation of COSMOS .COO and .COD file content. This allows maximum use of existing COSMOS routines for data set-up etc. The integration should provide all capabilities and options for the COSMOS back end from the start; only a limited set with the most important capabilities should be presented in the CCPN interface to start with.
    • .COO contains coordinates and all data generated by COSMOS. All data are assigned, which means that any ambiguous assignments have been disambiguated before the .COO is created. There are various strategies, to deal with this, including creating different COSMOS projects with different assignments.
    • .COD contains input data including NMR data, that can be unassigned, and information like charge group definitions, which atoms are backbone etc. (as regular exprressions). These should be read in two rounds to start with, first for NMR data, and second for standard setup information, like charge groups. Initially we shall keep a few standard .COD files for standard situations. We can get fully integrated later.
    • project files (.COS) contain switches and parameters. These will be read from CCPN and passed on to .COO or .COD. The format is keyword/value. For proteins most take default values.
    • Karlsruhe will set up their code to input/output data at the approppriate level and send specifications and documentation to Cambridge. Cambridge will look at sending data to and from these interfaces.
    • Karlsruhe will provide use cases and examples of COSMOS use, manuals and documentation for COSMOS back end, installations that work for 64 bit Linux, and will organise passing data to COSMOS front end using the same interfaces  as the data transfer to the back end.
    • CCPN will provide the standard NmrCalc / WmsProtocol interface with widgets etc. that is anyway being developed and provide WmsProtocol files for COSMOS to set up the interface. Together this yields a functional graphical user interface for  COSMOS. CCPN will expand the data model to deal with partial charges, and chemical shift tensors, and provide the code to pass  data to the COSMOS interface and call COSMOS from Analysis.
  • Development and use of web services will be dealt with after the first installation is completed in July. The web services will be designed in Karlsruhe, and the code to call them in Cambridge, as part of the Wms work (John Ionides). Karlsruhe will contact the WeNMR project (Alexandre Bonvin, Utrecht) for future coordination.
  • Code for transferring data between CCPN data model and COSMOS will be released as open source under LGPL license, as will code directly relevant to the CCPN data model. Release conditions for graphical user interface code are still to  be discussed.