Discussion forum for collaboration and formats.
Group review of data items and data model. Group report
on discussion and suggesting tasks required and collaborations
to be set up and mechanism for working together
A variety of software packages exist for performing
single particle reconstructions. For reference, the
three most widely used packages for single particle
reconstruction are:
Spider (www.wadsworth.org/spider_doc/spider/docs/master.html)
J. Frank, M. Radermacher, P. Penczek, J. Zhu, Y. Li,
M. Ladjadj, and A. Leith. SPIDER and WEB: Processing
and visualization of images in 3D electron microscopy
and related fields. J Struct. Biol., 116(1996),190-199.
Imagic (/www.imagescience.de/)
M. Van Heel, G. Harauz, E. Orlova, R. Schmidt &
M. Schatz. A new generation of the IMAGIC image processing
system. J Struct.Biol.116(1996), 17-24.
EMAN
(ncmi.bcm.tmc.edu/~stevel/EMAN/doc/) S. J. Ludtke,
P. R. Baldwin, and W. Chiu. EMAN: Serniautomated Software
for High-Resolution Single Particle Reconstructions.
J Struct.Biol. 128(1999), 82-97.
All three packages provide the necessary tools to
perform a complete 3D reconstruction from scratch, and
while significant differences do exist, all three packages
use conceptually similar techniques and cover the four
stages for structure determination: particle selection,
initial model generation, model refinement, and evaluation/resolution
determination.
It was also pointed out that other packages such as
the Image
Processing Toolbox Project (www.mathworks.com/products/image/)
may have some application in EM standardisation.
see also software listed under
em-outreach.sdsc.edu/community-codes/MicroscopySoftware.html
GRIP (GRoningen Image Processing).
MRC program package
Image-cnv.
XwindowsDisplay
Digital Micrograph, from Gatan.
GRACE a plug-in to Digital Micrograph
Topics Discussed
and recommendations for additional data definitions
to the EMDEP meta data
Filtering and image contrast Model accuracy
depends on a complete and quantitative accounting of
the image contrast. This should include not only single
elastic scattering events, but also inelastic scattering
and, if significant, multiple scattering. Furthermore,
the solvent and support film should be taken into account,
including contrast originating from rough ice and carbon
surfaces.
Measuring resolution: The FSC would be an unbiased
resolution measurement if the two reconstructions calculated
from two half data sets were truly independent, i.e.
if two references were used to refine parameters in
each set. The disadvantage is the reduced signal-to-noise
ratio of the two references as they would each contain
only half the data. An alternative would be the exclusion
of a certain percentage of data, say 5 or 10%, from
the refinement-reconstruction cycle. A correlation coefficient
or phase residual in resolution zones could then be
calculated between the excluded images and the reconstruction
to measure resolution. This idea, which is already implemented
in the program FREALIGN.
The particular aspects of using FSC for dividing data
at higher frequencies, for Tomography resolution should
be expressed in 3 dimensions, the same for 2d crystals
ditto. It was suggested that FSC should always be given
in three dimensions to give an indication of the effects
of incomplete sample /conical tilt.
CTF correction: the precise parameters have
to be determined not only for each image, but for each
particle in an image since the vertical particle positions
are all different. At high resolution, the curvature
of the Ewald sphere also has to be taken into account.
Apart from the defocus amount and astigmatism (3 parameters),
the beam tilt also has to be determined for each image
(2 parameters).
Beam tilt The effects on resolution from the
additional degrees of freedom due to parameters such
as magnification, defocus/astigmatism, and beam tilt
need to be recorded and considered.
Other factors discussed included
Time blotting/type
Distribution of orientations with asu
Characterisation of distribution of
defocus used
File compression methods
Interchange conventions template for
describing map Format/axis/directions etc and a request
for each package to make their details publiclly available
inc checksum.
Suggestion that the EMDEP system allow
for the archive of key stages from Raw to processed
data
Recording of the de-noise method algorthm
used pathway used to produce processed map
Segmentation e.g. via eigen vectors
auto characterised to be archived
Scan conditions and scanner type OD
transmission scaling including loss of data scaling
16 to 8bit
Density fitting and Domains movement
parameters
Fitting validation and Validation for
2d crystals data
Conventions for data exchange and archiving
the same, including standard for symmetry axes
Map conventions
image standards
Actions to
be taken by EMDEP
1 EBI create web page for registration
of new data items
2 use 3dem email list as active
discussion medium
3 letter to editor expressing this
meetings agreement
4. Possible add category of data
items to show process steps for raw data to processed
data (+ params at each step). Allows for re-submission
of same original data as new data set
5. At this stage limit deposition
to processed maps and be prepared when band width is
there to add more exp data/raw data when physically
possible?
6. Local disk storage of exp data
– corresponding to deposited data/id
7. FSC yes curve as table of values
to de deposited + supporting data kept at least locally
where applicable highly recommended
8. Suggest deposition of masks/data
halves for FSC but not mandatory at present
9. Full description of biochemistry,
cell biology of sample e.g.
- protein localisation - cell and
life cycle stage for sample - Tissue cell organelles -
anatomical details - region of interest - species - system
- organ - region - cell type - structure
Some of these characteristics are defined now
e.g.
natSrcType
element name="organism"
element name="synonyms"
element name="cell"
element name="organ"
element name="tissue"
element name="organelle"
element name="secretion"
element name="details"
engSrcType
element name"organism"
element name="host_organism"
element name="host_culture_collection"
element name="host_vector"
element name="host_vector_type"
element name="details"
10. EMDEP system to expand the related
database links from the current set
nomclType
name="ref_go"
name="ref_interpro"
name="ref_ictvdb"
Possible links may include disease related IDs
and the Cellular Properties Database (CellPropDB),
the OdorMapDB , and SenseLab OrDB
11. EMDEP to look into aspects of the
data model for Multi techniques, Multi resolution, Techniques
3d/4d
|
