Visualization (led by Joachim Frank)
Superimposition of structure and density using semi-transparent density –
Flexible fitting of X-ray structures into EM maps – complex views
Interpretive diagrams highlighting specific structural features with
o Quick view – essential views with simple functionality
o Different display modes: rotating display, surface rendered,
contouring, volume-rendered and slices
o Low-pass filter for comparison of structure and density maps with
o Segmented volumes volume by 3D masks with multiple views and
o Interactive modes of viewing – steer viewing direction with angles
reported; dial in angles to define viewing direction
o 3D maps: variance map provides visual display of uncertainty per
voxel – Provide joint viewing with structure and other views above
with flexible navigation
Are special tools required to address the visualization of structure at
different levels of resolution?
Handling of transparency in existing X-ray map visualization is problematic
for view EM map data? Part of this is a problem of incorporating new
technology into existing codes. PMV (M. Sannner) and tools by C. Bajaj
are examples of packages that exploit the new technologies.
The importance of volume and surface rendering is more central to the
reporting of results of EM and less demanding in X-ray.
Commercial versus public packages (public packages = vmd, pmv/viper,
chimera)? Lots of public software components exist but may need to be
adapted to biological and EM visualization applications.
Wikipedia is an example of on-line annotation system
Need tools that can be used by novice users. The tools provided by public
archives must be simple and very accessible to a broad audience. Other
tools will be required for research applications. This may be resolved by
adapting user interfaces.
Applet versus application? What requirements should be placed on the
user to do visualization? In the case of the PDB, visualization should not
require additional software installation for the novice user.
Generation of reports, charts, and histograms
Generation of spreadsheets and charts to view important parameters of
o Refinement history and defocus distributions
o Classification dendograms - visual depiction of particle coverage
o Angular distributions
Some of this may be pre-computed at deposition time.
Standard reports across all structures would be useful.
Future activity - wait
3D motif search and automated docking
o with user supplied motif targets
o with motifs based on PDB structures
o with motifs defined on existing density maps
Other disciplines are working on related search and matching problems
and often developing tools in the abstract.
What are existing tools for mining and display?
Data mining (Led by Andreas Engel)
Goals and examples
o Atomic scale model of cells (ribosome example)
o Atomic models of interacting complexes
Actin mesh & aldolase segmentation example Volkman 2002
Helix Hunder – Wah Chiu @ ~ 10 Angstroms
Beta-sheets and Helix orientation at @ ~5 Angstroms De-
Groot J. Mol Biol (2000); Mitsuoka, J. Struct Bio 1999)
HBV example - Fitting of FAB - Alasdair Steven
o Protein motion
Decompoistion of continuous object into a set of Voronoi
cells that approx. the shape of the density distribution. The
resulting elastic network allows themal fluctuations to be
analyzedMing, PNAS 2002 99, 8620-25; Tama, J. Mol Biol.
(2002) 321: 297-305, Scheuring, E. Biophys J . (2001)
Atomic force microscope example (movie of ~1500 images
first determining an average followed by the sequential
display of spatially correlated images)
What are the data mining capabilities for X-ray structures? What is the
role of the archival database in developing mining and analysis tools.
Links provided to boutique databases, 3rd party annotation and analysis.
The division of effort between the RCSB PDB archival database and
related research projects was described.
Access should be provided to permit community to easily contribute to
features of the archive.
Are AFM structures appropriate for archival deposition? Would require a
data dictionary of terms.
Data Integration (Led by Helen Saibil)
Integration of MAP data into PDB
o Higher visibility of EM database
o Integration with biological data –
o Integration facilitates training
Role model – protein crystallography
o CCP4 model collaboration – software tools and training –
Basic features of EMDB
o Front page identity
o Search on entry number, author, keyword
o A friendly data path with a picture of the map
o Bug-free map header
o Links to sequence and function databases
o Display of experimental information
o Validation information
o Links to and from PDB
Desirable features for EMDB
o Rasmol for maps
o Tools for moving and aligning maps - superposition
o Map and structure alignment – shift map origin
o Searching densities by cross correlation
o Integrate with cell structure data
o Validation with Rosenthal/Henderson tilt data (experimental cross-
o Fold search for maps at < 7Angstroms
o Full integration into the PDB
What is the scope of integration among other EM techniques?
How should the integrated resource be named?
Unification of resources is important both data and query. Don’t duplicate the
experience of other communities (e.g. astronomy).
Still need a single interface to query and retrieve structure data from all
structure techniques including EM.
Gordon Conference for EM may be a good venue for continuing these
discussions. Video conferencing may be another alternative.
PyMol or ASTEXViewer may be useful tools for map display and integration
WRAP-UP – (Led by Helen Berman)
Workshop results on the web
Workshop will be summarized in up-coming article –
Revise dictionary and update data sets
Develop plan for RCSB PDB and EM/MSD integration
Coordinate publication requirements with deposition requirements – and the
journal requirements may vary with resolution …
o Any publication containing a description of atomic coordinates should
be accompanied by a PDB deposition. The policy with respect to the
deposition of maps should be defined.
o Letters to relevant Journals should be drafted advocating this policy.
o How should PDB validation be modified for low-resolution structures?
o Should rigid body refinements be included in PDB?
o How should the resolution/reliability of the model be conveyed to the
o Release policy – funding agencies want to have a role. Draft a policy
to be circulated among the community, journals and funding agencies
modeled after IUCr policy. Should the release policy mandate the
some criteria for resolution determination? Make the content and
handling of date explicit in the release policy.