[3dem] relion-1.4 release
Sjors Scheres
scheres at mrc-lmb.cam.ac.uk
Wed Sep 2 01:52:29 PDT 2015
Dear colleagues,
Please be informed that relion-1.4 is now available for download. It's
main new features are:
1. SUB-TOMOGRAM AVERAGING
As published in our latest Structure paper
(http://dx.doi.org/10.1016/j.str.2015.06.026). Detailed instructions are
avaliable on the Wiki. Apart from the concept (and calculation) of a 3D
CTF-model, for which Tanmay Bharat wrote some scripts, most things are
similar to single-particle processing. So, if you hadn't used relion for
single-particle processing and wish to perform sub-tomogram averaging,
please do browse through the Wiki pages, and perhaps follow the tutorial
first.
2. FOCUSED CLASSIFICATION WITH SIGNAL SUBTRACTION
Technically not a new feature, because the main functionality was
already present in relion-1.3 (though user-friendliness has been
improved). This new tool is probably our most powerful classification
approach yet. We demonstrate its results on our gamma-secretase dataset
that gave a 3.4A map, in which we now find three sub-states. The
corresponding paper has only just been submitted for publication, but to
facilitate early uptake of the method we have put a preprint on bioRxiv
(http://dx.doi.org/10.1101/025890). Detailed instructions on how to use
this method may be found in the preprint.
3. SPEED-UP
a) Re-implementation of multi-threaded code. Now, many-threaded jobs
(e.g. --j > 6) scale better than before.
b) Movie-processing is now >10x faster than before.
c) By default, relion_refine_mpi will now read images from ALL
MPI-slaves in parallel. This was only done by the MPI-master in
relion-1.3. This will speed-up significantly on parallel file systems
(we use /fhgfs). However, be careful with large NFS-based system, which
may not be able to cope with the parallel load. You can go back to
master-only reading using the --no_parallel_disc_io flag.
4. MEMORY USAGE
There is now an option (in the INSTALL.sh script) to compile the entire
package in single-precision (the default remains double-precision).
Reproducibility of these runs is gone, but it saves you 50% RAM. Our
results indicate that performance does not suffer significantly, but
admittedly not as many tests were performed with single-precision mode
as with the other new features. Therefore, your feedback on this feature
will be much appreciated.
For those of you who participated in beta-testing: thank you for your
help in making the code more stable! Please update your
beta-installation to the stable release.
Have fun,
Sjors
--
Sjors Scheres
MRC Laboratory of Molecular Biology
Francis Crick Avenue, Cambridge Biomedical Campus
Cambridge CB2 0QH, U.K.
tel: +44 (0)1223 267061
http://www2.mrc-lmb.cam.ac.uk/groups/scheres
More information about the 3dem
mailing list