Transcript Kappa-05-2008 - EPN

Status report



Data model for goniometry and collision
maps
CCPN-generated Java API (file-based
backend)
Used by:


GΦL
STAC: to persist a pure-rotation description of
the goniostat and a set of collision maps for a
range of detector distances
BEST: to calculate a set of possible strategies
exploiting Kappa geometry
Leiden, May 2008
The future


Further testing and development
Extension of scope of model




All beamline instrumentation, not just goniostat
Output of strategy calculations, and data
collection
Perform real data collections according to the
calculated strategies
Collaboration between Global Phasing and
Soleil
GΦL
Leiden, May 2008
People involved

Sándor Brockhauser


Johan Unge, Gleb Bourenkov


EMBL Grenoble
EMBL Hamburg
CCPN: Rasmus Fogh, Wayne Boucher,
Ernest Laue

GΦL
Biochemistry Department, Cambridge
Leiden, May 2008
GΦL
Leiden, May 2008
BioXDM coordinate systems

Described by:




The matrix required to convert from a canonical
coordinate system
The incident beam direction
The component of the principal goniostat axis
perpendicular to the beam
An orientation vector/matrix is:

GΦL
A unit vector/orthogonal matrix with an
associated coordinate system
Leiden, May 2008
BioXDM goniometry model


Completely general multi-axis (not Kappa
specific)
Pure rotation



GΦL
Each axis represented as direction (given by a
unit vector) + setting
These directions may change from one data
collection to another (e.g. By re-calibration)
No account taken of physical motors or
translations
Leiden, May 2008
BioXDM collision map

The “axes” of the collision map are identified
with “axes” of the goniostat


A set of collision maps are stored, over
several detector distances


In the Kappa case, Omega and Kappa
A utility function provides an appropriate collision
map for a specified detector distance
Each point on the map is an object in its own
right
GΦL
Leiden, May 2008
GΦL
Leiden, May 2008
GΦL
Leiden, May 2008
GΦL
Leiden, May 2008
BioXDM and data modelling

Why do data modelling? Reasons include:


Emphasis on science first, then implementation
In the BioXDM context, provides “glue” to
connect diverse applications together

GΦL
c.f. BEST and STAC
Leiden, May 2008
BioXDM and CCPN

Currently use the CCPN approach


GΦL
Python and Java API's are generated from the
data model (with a small amount of hand-written
code)
API's perform a lot of data integrity checking
(derived from data model) that would be tedious
and impractical to code by hand.
Leiden, May 2008
CCPN status



Java API now able to use an SQL backend
via Hibernate (at least in prototype form)
Python API only supports XML file backend
Rely on ObjectDomain for data modelling

GΦL
this is an unsupported product
Leiden, May 2008