Announcements related to binary/source releases

1436 views
Skip to first unread message

Steve Ludtke

unread,
Aug 5, 2011, 10:50:32 AM8/5/11
to em...@googlegroups.com
This discussion will cover temporary availability issues and new release announcements

Steve Ludtke

unread,
Nov 18, 2011, 11:44:44 AM11/18/11
to em...@googlegroups.com
Announcing EMAN2.04

This version includes a substantial number of new features over the 2.03 release (though all have been available through the nightly snapshots).
Major improvements include:

- New e2filtertool.py program for interactive filtering and image processing of both 2-D and 3-D data
- New e2imageval.py program for interactive power spectrum analysis of raw (unboxed) CCD frames and scanned micrographs. Provides a variety of tools for subregion selection and basic CTF analysis. Designed for interactive use on the microscope during data collection, or to rapidly screen sets of images for quality.
- New program examples/mergeproject.py which will take 2 projects and merge them into a single new project. For large projects with data in multiple groups, permits the groups to be processed independently, then merged together at some point for a large-scale refinement.
- New processor 'math.simulatectf', which in conjunction with e2filtertool.py can be used to prepare very realistic simulated particle images.
- Improvements for support for icosahedral structures. Better tools now provided for moving between the various symmetry axis conventions.
- e2ctf.py now shows a subset of the particle images associated with each data set
- Color display of Fourier phases has been fixed.
- MPI parallelism improvements, including experimental new option for disabling image caching for clusters with distributed filesystems like Lustre..

Steve Ludtke

unread,
May 10, 2012, 1:32:35 PM5/10/12
to em...@googlegroups.com
Announcing EMAN2/SPARX  2.05
(OSX binaries will be posted later today, others available now)

EMAN2 changes include:
  • A fix for the problem impacting certain projects where tasks would crash with "unable to allocate memory for mutex" errors (please report it if you still have issues with this)
  • New "gold standard" resolution test: e2refine_evenodd.py. Expensive, but not subject to noise-bias.
  • Full support for performing tilt-validation of structures (Rossenthal & Henderson method)
  • Drag/drop curves between 2-D plotting windows (useful for making CTF comparison plots)
  • Preview version of replacement for e2workflow.py: e2projectmanager.py. Not yet 100% complete, but far less reliant on the database system.
  • New file browser and new 3-D display widget available through e2projectmanager.py (folder button)
  • Improvements to Single Particle Tomography (subtomogram averaging)
  • Many bugfixes

SPARX now includes
  • a full implementation of codimensional PCA (described in Penczek, P. A., Kimmel, M., and Spahn, C. M. (2011) Identifying conformational states of macromolecules by eigen-analysis of resampled cryo-EM images, Structure 19, 1582-1590.) including a full demo with simulated data.
  • implementation of the new 2D method ISAC (see Yang, Z., Fang, J., Chittuluru, J., Asturias, F. J., and Penczek, P. A. (2012) Iterative stable alignment and clustering of 2D TEM images, Structure 20, 237-247.)

Steve Ludtke

unread,
May 24, 2012, 10:47:54 AM5/24/12
to em...@googlegroups.com
We have just discovered a serious problem with the Windows binaries. Some floating point numbers appear not to successfully go from Python to Windows at the moment. It is a very odd bug, as some specific numbers are converted properly, but others aren't. We are sorting it all out now. For the moment the 2.05 Windows binaries have been removed. I will post again here when a corrected version is available. The Linux and Mac binaries are fine.

Steve Ludtke

unread,
Jul 25, 2012, 10:19:28 PM7/25/12
to em...@googlegroups.com
EMAN2 on Mountain Lion (Mac OS 10.8)

Please be advised that at present, EMAN2 will not work on MacOS 10.8. Apple has removed several libraries from the system that EMAN2 relies on. We are working on a solution, but it may be a week or two before we have new binaries in place.

Steve Ludtke

unread,
Aug 1, 2012, 9:00:53 AM8/1/12
to em...@googlegroups.com
In theory, there is now a working MacOS 10.8 (Mountain Lion) version, and it is possible to compile on 10.8. Please report any problems you have.

Steve Ludtke

unread,
Aug 1, 2012, 9:22:55 AM8/1/12
to em...@googlegroups.com
I realize this is a somewhat unorthodox place to post this, but I'm trying to cast as wide a net as possible. Grant Tang, currently the only professional programmer I have working on EMAN2, is moving on to a new job opportunity he couldn't refuse, and I am now seeking a replacement. I'm posting a brief description of the position here. Please feel free to circulate if you know any appropriate candidates.

----
Seeking an entry level programmer for development of the EMAN2 software suite (http://blake.bcm.edu). EMAN2 is a large scientific image processing suite with cross-platform support (Mac, Linux, Win). The position involves  a range of different skills, including OpenGL, C++, Python, cluster computing, CUDA, image processing and cross-platform compilation. Experience in all of these areas is not necessary, and we are happy to provide promising young programmers with on-the-job training, increasing  your future marketability. In addition to software development, the position will involve maintaining a simple automatic cross-platform compilation system, and occasional email-based user-support. Please contact sludtke <at> bcm.edu directly with a CV/Resume or with any questions.

Steve Ludtke

unread,
Jun 15, 2013, 1:35:03 PM6/15/13
to em...@googlegroups.com
Announcing EMAN2/SPARX 2.07

We are pleased to announce the availability of EMAN2/SPARX 2.07. There are many minor usability improvements since EMAN2.06, but the main reason for this release is that the nightly snapshot versions will transition to the EMAN2.1 prerelease branch within the next week or two. EMAN2.1 has been a major effort for the last 2 years, and includes MANY new improvements (some of which have appeared in EMAN2.0x already). However, to use EMAN2.1 you will have to convert EMAN2.0x projects to the new (non-BDB) standards. That is, there will be a period of a couple of months where the nightly snapshots may not be 100% functional, and you will not be able to easily switch from a 2.1 release back to a 2.0 release. 2.07 thus represents the final stable release (we hope) in the 2.0x series.

We will release more details about 2.1 and how to transition gracefully over the next few weeks, but I am 100% confident that you will be happier with 2.1 than 2.0. The two major improvements you'll see are the retirement (from e2* programs) of BDB and the addition of e2refine_easy.py which has only a few parameters, performs integrated gold-standard resolution evaluation, and is a minimum of 2x faster than the current programs.

Some of the major changes since 2.06:
  • 'Gold Standard' resolution test (e2refine_evenodd) improved, and should be used for all structures (in 2.1 it will be integrated with refinement).
  • Uses the new unified EMAN2/SPARX MPI installer
  • Histogram equalization in 2D image display for improved contrast
  • New 'probe' mode in 2D image display
  • Keyboard shortcuts in e2ctf.py GUI for rapid screening of large projects
  • Improved DM3 support (Summit) and DM4 support added. Other file format compatibility also improved
  • Many more...

Steve Ludtke

unread,
Jul 2, 2013, 7:48:31 PM7/2/13
to em...@googlegroups.com
The first pre-release version of EMAN2.1 is now available for download (EMAN2.1alpha1). This version is not 100% stable, and is not complete, but the full pipeline for standard single particle reconstructions should be fully functional. That is, if you're anxious to play with a non-BDB EMAN2, and want to see how the new tools perform, you can download it and give it a try, but there will almost certainly be significant bugfixes and features added over coming weeks. Some of these changes may make projects refined with 2.1alpha1 not work properly. That is, it's is an experimental version, not yet ready for routine use.

Details are available here: http://blake.bcm.edu/emanwiki/EMAN2/Eman21Release

Bug reports are, of course, welcome.

Please also note that this version does not include the most recent changes to SPARX, so for heavy SPARX users, it may be better to wait for alpha2, which will merge all recent changes together.

Steve Ludtke

unread,
Jul 3, 2013, 1:27:23 AM7/3/13
to em...@googlegroups.com
Much of the Wiki is in a bit of disarray at the moment as we begin to update the documentation to reflect changes in 2.1 (like removing all of the BDB warning messages). Anyway, I just finished manually updating the EMAN1->2 transition guide quickstart so it reflects all the changes in 2.1:

http://blake.bcm.edu/emanwiki/EMAN2/Eman1Transition/QuickStart

While this is hardly a full, detailed, tutorial, it should be enough to get experienced cryoEM people started, and I believe it's fully up to date now.

Steve Ludtke

unread,
Jul 6, 2013, 2:11:17 AM7/6/13
to em...@googlegroups.com
A fully updated, and somewhat expanded tutorial covering EMAN2.1 is now available. This was prepared for the cryoEM workshop in Beijing next week. Following this tutorial, using EMAN2.1 with a quad-core laptop, an experienced user can go from raw micrographs to an ~8Å resolution GroEL structure in ~3 hours, including both manual and computer processing steps. (http://blake.bcm.edu/emanwiki/Ws2013/EMAN2)

Please do let me know if you run into problems with the 2.1alpha1 release crashing or experiencing other problems not listed on the release page (http://blake.bcm.edu/emanwiki/EMAN2/Eman21Release).  If you go through the tutorial and run into issues with it, please also let me know so I can update it.

cheers

Steve Ludtke

unread,
Jul 7, 2013, 5:47:06 AM7/7/13
to em...@googlegroups.com
Just a quick note that the Linux binaries of 2.1alpha1 were just updated again silently (they still look like they are from 7/2/13, but are actually from 7/7/13). The previously posted version would work with current Linux distros, but was not compatible with older CentOS releases running on some clusters. The new version should be more backwards compatible (though there are limits). Thanks Ian !

Steve Ludtke

unread,
Jul 23, 2013, 10:34:35 PM7/23/13
to em...@googlegroups.com
We're making good progress towards the 2.1 release. The 2.1 alpha release could be used to run full refinements, but many other features were not yet ported, or were broken. The nightly snapshots are now on the 2.1 branch rather than the 2.0 branch, so if you are experimenting with 2.1, please try the nightly snapshot to make sure the problem hasn't been fixed already.  Alpha2 should be out within a week:

Changes since the 2.1alpha1 release:
- In e2refine2relion3d, the defocus values were supposed to be in angstroms, but were passed in nm instead (there WAS a reason this happened). In any case, EMAN2-based Relion runs should provide MUCH better results now.
- Progress monitor in e2projectmanager was broken, and causing incorrect entries in .eman2log.txt. Works properly now (linux and mac)
- occasional problems with MPI parallelism resolved (I think)
- e2boxer fixed
- fixed alignment of even/odd maps in e2refine_easy to take symmetry into account
- fix problem with e2refine2d  always using the same output directory
- added --prethreshold option which dramatically reduces noise in maps when projections are made

Known issues still being worked on:
- e2refinemulti not ported yet
- e2evalparticles only partially functional
- e2spt pipeline not completely ported
- Frealign conversion still being ported
- New HDF pointer caching mechanism still in testing
- e2refinevariance not ported yet
- RCT and tilt validation needs to be tested, tilt .box file import broken
-

Steve Ludtke

unread,
Oct 23, 2013, 9:29:46 AM10/23/13
to em...@googlegroups.com
For Mac users out there, we have tried OSX Mavericks on several computers. From Mountain Lion it is a painless (and completely free) upgrade, and we have not experienced any problems.

As far as we've observed so far, the current EMAN2/SPARX releases for Mountain Lion (10.8) should be fully functional on Mavericks (10.9).

For users still lingering on Snow Leopard (10.6), just a reminder that we will not be supporting 10.6 in any releases after 2.0.7. Given that this is a free upgrade, and can supposedly be done in a single step from 10.6.x, I would strongly encourage users to consider it. Of course, you will have to contend with 3+ years of improvements to the OS all at once, and I cannot speak for any other software you have, but eventually you are likely to have to bite the bullet.

If you experience any problems, please report them !

cheers

Steve Ludtke

unread,
Dec 10, 2013, 12:28:17 PM12/10/13
to em...@googlegroups.com
Just a quick warning. Over the last week or so we were making some improvements to the masking system, so it was possible to apply a soft inner-mask to virus particles after each refinement iteration. If you downloaded a daily release over this timeframe and are experiencing any unexpected errors related to post-processing or masking in e2refine_easy, please download a new daily release. I believe all of the problems have been fixed now.

BTW - if you need inner-masking capabilities, there is now as mask.soft processor which permits both inner and outer masking with a specified Gaussian width. To do this, specify the :  --automask3d2 option with whatever additional mask you'd like to apply. This mask will be applied after the normal automask, so, for example, you could apply a soft Gaussian on the inside of the capsid, but still use the auto-mask for the outside (by specifying a very large outer radius for mask.soft), to reduce the chance of resolution inflation.  Whenever you use any sort of fixed mask, even if it is 'soft' you need to be wary of model bias and the resulting resolution exaggeration.

Steve Ludtke

unread,
Feb 28, 2014, 12:22:50 AM2/28/14
to em...@googlegroups.com
Please note that the Wiki is offline due to some significant spam issues. It should be back up within ~24 hours.  Sorry for the inconvenience.

Steve Ludtke

unread,
Mar 5, 2014, 4:03:42 PM3/5/14
to em...@googlegroups.com
The Wiki is back online again. Please report any problems (other than the new appearance).  If you are interested in contributing edits to the Wiki, shoot me an email and I will give you an account . We aren't trying to exclude anyone here (except for the !@#$ spammers).  The appearance will probably continue to change over the next couple of days. If you preferred something more like the old appearance, users with accounts can set the appearance in their preferences.  Thanks for your patience while we eliminated ~90,000 spam pages...  ugh

Steve Ludtke

unread,
May 27, 2014, 7:21:59 PM5/27/14
to em...@googlegroups.com
We are pleased to announce the release of the first Beta version of EMAN2.1 available for download now.  While there are likely still a few bugs hovering around this version has been very stable in in-house testing. We believe it is now a good time for anyone still hanging on to 2.0x to make the switch to EMAN2.1. There are many good reasons to do so. Please report any bugs or problems you encounter either on this mailing list or via direct email to sludtke -at- bcm.edu.

http://ncmi.bcm.edu/ncmi/software/counter_222/software_125

Important changes since the 2.1 alpha release:
* e2refine_easy much more efficient than EMAN2.0x, and produces gold-standard FSCs on the fly (~20x faster than Relion in testing)
* e2refine_easy automatically selects almost all parameters (and explains its selections)
* significant improvements to refinement reports
* tiltboxer, RCT and tilt-validation should all be working now
* evalparticles fixed
* Full support for EMAN2 -> Relion, Relion -> EMAN2 and EMAN2 -> Frealign (more coming)
* e2projectupdate21 to convert EMAN2.0 projects to 2.1
* MPI support has been completely revamped for more efficient cluster computing
* Volume series support in file viewer
* Dramatically expanded single particle tomography (subtomogram averaging)
* Revamped CTF fitting, including astigmatism support
* Experimental support for direct detector movie-mode processing

Steve Ludtke

unread,
May 27, 2014, 9:21:57 PM5/27/14
to em...@googlegroups.com
Aside from the fact that I clearly need more sleep based on the grammatical issues in the original post, I wanted to add that the Windows installer for this version will claim to be EMAN2.06 when you run it, though the date will be correct. After installing, if you run e2version.py you should see that you did, indeed get the correct version.

Also worth pointing out that the binaries for all three platforms focus on compatibility rather than performance (the Linux binary will run on 7-8 year old machines, and the Windows 32 bit binaries will probably run under XP). On my test computer if I compile EMAN2 myself with ENABLE_OPTIMIZE_MACHINE, I get an e2speedtest of ~1.05. With the current 64 bit binaries, this falls to ~0.6-0.7 and on Windows 32-bit ~0.4. That is, a refinement with an optimized Linux compile will be fully 2.5x faster than running on Windows, a not insignificant factor.  We will probably try to put out a more aggressively tuned binary option for linux sometime soon, but it may be difficult to do much for Windows. That is, the Windows binaries are really mostly for interactive, graphical tools rather than real refinement work...  We may give Cygwin a try and see how that fares.

Steve Ludtke

unread,
May 30, 2014, 10:50:35 AM5/30/14
to em...@googlegroups.com
Ok, as feared, some little glitches in 2.1beta1. In particular there is an error which will cause e2refine_easy jobs to crash after completing 1 iteration.  We will be posting 2.1beta2 sometime in the next few hours...

Steve Ludtke

unread,
Jun 13, 2014, 1:17:06 PM6/13/14
to em...@googlegroups.com
(sorry if you also got this via 3DEM)

I decided to group three related announcements into a single message to limit the damage to people's inboxes :^)

---
First, a request to the community. The R01 grant supporting EMAN maintenance and development is going in for renewal in three weeks. I have a brief survey set up here (hopefully you don't have to type that URL by hand):

https://docs.google.com/forms/d/1vtrgk0uZ4Ycgw0LZEWPkUwJ28IjVb6TiNtuH-AxWWRQ/viewform?usp=send_form

If you've ever used EMAN, I'd greatly appreciate it if you could take a couple of minutes to fill it out. It really isn't very long.

---
Second, since we haven't managed to host our big week-long workshop in a couple of years due to site/timing issues, we've started running (mostly) weekly live telecast tutorials on a range of topics. These are then archived as informal video tutorials. The first of these, dubbed:

EMAN2.1 - One hour to subnanometer resolution (on a laptop)

is available online via the EMAN wiki http://blake.bcm.edu/emanwiki/EMAN2/VideoTutorials and via my (sludtke42) YouTube channel.
---

Finally, after over a year of development, we are very pleased to announce the availability of EMAN2.1 (beta3).

http://blake.bcm.edu/emanwiki/EMAN2

While we haven't quite put the "release" moniker on it yet, the beta3 is very stable and ready for day-to-day use.  Major improvements since EMAN2.0x include:

* Retirement of the "BDB" system which drove so many users crazy
  * All images are now stored by default as flat HDF files (chimera compatible)
  * Metadata is stored in human-readable JSON files, which can be seamlessly copied between projects
  * Easy e2projectupdate21 script to move from EMAN 2.0

* e2refine-easy
  * True Gold-Standard FSCs
  * Automatic FSC-based map filtration and masking
  * Automatic selection of almost all refinement parameters based on your data and built-in heuristics
  * Much faster due to automatic parameter optimization (typically 2-10x)
  * Generates full web-based refinement report including resolution plots, updated live during refinement

* Intersoftware compatibility
  * Convert EMAN2 projects in one step for cross-testing in Relion and/or Frealign
  * One-step Relion project conversion to EMAN2
  * EMX support (this is the future !)

* New graphical tools including multi-model 3-D visualization with sequence animation, Slices and annotation

* Revamped MPI support for compatibility with more clusters and interoperability with (co-distributed) SPARX

* A full suite of Single Particle Tomography (subtomogram averaging) routines, with a wide variety of available methods and GUI tools

* Improvements to file format support, including compressed TIFFs (often used with K2 movies), DM4, Falcon and others


If you are an EMAN2.0 user, I strongly encourage you to take the leap and upgrade now. For users of other software, we are trying very hard to make it easy to move between packages for comparative purposes.

We have done a number of validation tests among software now, and for "good" data sets, you can achieve virtually identical results using Frealign, Relion and EMAN2 (and almost certainly others). However, for heterogeneous or flexible molecules, you will find differences, making cross comparisons among software a valuable test in validating your results, or at least in determining how far you can reasonably trust them.

cheers

Steve Ludtke

unread,
Nov 14, 2014, 2:34:08 AM11/14/14
to em...@googlegroups.com
We are pleased to announce the official release of EMAN 2.1.  EMAN2.1 includes many improvements over EMAN2.0x. If you haven't switched to one of the 2.1 pre-releases it is time to upgrade. We have tested refinements in this release version quite extensively, and everything is working well. There are still a few improvements we hoped to make which didn't make it into the 2.1 release, like improvements to the particle picking tool, but most of the major changes are complete and stable in this version. I'm announcing this version on the EMAN2 mailing list first, in the hopes that more active users will download and test it. If no major problems arise, I will announce it on the 3DEM newsgroup and via other venues.

Please report any problems you encounter!

Note that I also intend to resume the bi-weekly live Google training sessions next Thursday. Details appear on the main Wiki page.

Steve Ludtke

unread,
Nov 19, 2014, 4:05:18 PM11/19/14
to em...@googlegroups.com
Now that my traveling is done for a while, I will be resuming the Thursday 'Google Hangout' sessions, starting tomorrow (11/19/14) morning at 10 AM. Quite a few new capabilities were added before the 2.1 release, including some much needed convenience tools, such as the ability to work with downsampled particles within the same project as the fully sampled data. I have a new tutorial data set we will make available sometime soon which allows you to get to better than 5 Å resolution in a couple of hours on a single workstation.  I will start going over this data, and explain how I approach a new project with the tools available in 2.1.

11/19/14 10 AM CST (same time zone as Chicago)

Just follow this link, or look for sludtke42 on google:
https://plus.google.com/events/c7jgmcogdrlgkjne5g0oqhs7qug

Live participants can type their questions during the session and have them answered (almost) live. The session will also be recorded for later viewing. Other than a Google account, nothing special is required to participate.

Steve Ludtke

unread,
Jun 5, 2015, 9:46:24 PM6/5/15
to em...@googlegroups.com, slud...@gmail.com
We are very pleased to announce the official release of EMAN2.11 (as always this also includes the current version of SPARX, though this list is EMAN-centric).

Direct download links are here: http://ncmi.bcm.edu/ncmi/software/counter_222/software_130

On this list, I will assume everyone is aware of (and hopefully using) EMAN2.1 or some snapshot release, so I will focus on the changes since the last official release:

New Features:
  • e2refinevariance working again
  • Work with downsampled particles in the same project as the originals to combmine speed with final high resolution refinement
  • e2proc3d on stacks of volumes now uses much less RAM and is much faster
  • Dramatically improved speed of subtomogram alignment, particularly for large particles (with correct options)
  • MRC particle stacks can now be read by any program directly as a stack of particles (must use .mrcs extension)
  • Minor resolution improvement in single particle reconstruction do to improvement in reconstruction algorithm
  • e2proc2d and e2proc3d options to select exact data output mode for all formats (float, int, byte, etc.)
  • Updated support for FEI SER format
  • Support for EMX data exchange format
  • Support for CTFFIND parameter import
  • Initial support for display of atoms in 3-D
  • New e2proclst program for creating and manipulating .lst files
  • Improved whole image CTF fitting in e2evalimage
  • Experimental per-particle CTF correction for subtomogram averaging

Significant Bugfixes:
  • Fixed a bug that occasionally caused some very high resolution noise to appear in 3-D reconstructions
  • Fixed problem causing strange edge effects to sometimes appear in unmasked reconstructions
  • Fixed a bug causing strange edge effects and poor centering in some 2-D class-averages.
  • Fixed a problem with marking bad particles manually.
  • Some fixes for CTF correction with astigmatism
  • New eman2.py program to reduce new user confusion

Steve Ludtke

unread,
Oct 21, 2015, 8:27:50 AM10/21/15
to EMAN2
We are pleased to announce the release of EMAN2.12  (which, as always, also includes the current version of SPARX).

This is a more significant release than 2.11 and includes a range of new features, not just bugfixes. The major changes include:
  • New subtomogram averaging programs, and a new subtomogram alignment routine which is orders of magnitude faster than the old routine for large volumes. It is now possible to perform subtomogram averaging on thousands of 300x300x300 subtomograms on a workstation.
  • The return of the --usefilt feature from EMAN1. The new e2refine_easy option is --inputavg, and permits using one set of particles for alignment, and a separate set (perhaps with lower dose in movie-mode imaging) for reconstruction.
  • New e2refine_split program which will split a  single map reconstruction with heterogeneity reconstruction into 2 submaps, identifying the largest heterogeneity present. This program supplements e2refinemulti.
  • e2initialmodel_hisym for high symmetry (icos,tet,cubic) initial model generation.
  • reorganization of the workflow and addition of many previously command-line only programs
  • direct support for MRCS 4-bit PACK mode in both of its forms
  • improvements to e2refine_easy heuristics for slight improvements to reconstruction quality
Of course there is the usual set of minor bugfixes and other changes.

One warning for Mac users. There are a number of 'issues' with running EMAN2.12 on ElCapitan (OSX 10.11). These problems are more annoyances rather than things that really prevent you from doing reconstructions, and largely focus on GUI issues. As a simple example, if you use the browser to display an image, then open the control panel for the image, when you close the image, the control panel window doesn't close immediately. These are not general EMAN 2.12 bugs, but are ElCapitan-specific, and may take some time to properly debug.

Steve Ludtke

unread,
Dec 15, 2015, 7:18:11 PM12/15/15
to EMAN2
We have just completed a transition of the official EMAN2/SPARX source repository from an internal CVS archive at BCM to  GitHub:

https://github.com/cryoem/eman2

For end users who normally download the binary releases, this change should not impact your interactions with the software at all at this point in time. In the future, we may also transition the binary distribution to a more "pure open-source" repository. We will continue to make new binary releases, and make them available using the same page we currently use.

For advanced users/programmers this changes the way you can interact with EMAN2. You can now download a snapshot of the source code at any point in time, and see a complete history of what has changed. This way you won't get "stuck" if we happen to have a bad daily snapshot release. You are also free, of course, to make your own branches of the source code, and if you come up with a suggestion for a change, you can submit a pull request to ask that it be included in the main EMAN2/SPARX branch.

This marks the first step in a series of planned upgrades to the EMAN2/SPARX infrastructure as we move forwards towards the next major release.

Steve Ludtke

unread,
Oct 26, 2016, 9:39:19 PM10/26/16
to EMAN2
As many of you probably aren't aware (unless you're a programmer), GCC, the C++ compiler used for EMAN and everything else on linux has several major versions around right now: GCC4, GCC5 and GCC6.  While it may seem that something like a C compiler version wouldn't likely have much effect, in a recent test on the exact same machine, I got:

e2speedtest.py:
GCC 4.8:  ~0.9
GCC 5.4:  ~1.3
GCC 6.2:  ~1.5

That is, compiling with gcc 6.2 made the numerical code in EMAN2 run over 1.5x faster on the exact same hardware.  Now, why is this important? Different linux distributions use different versions of GCC.  In some of these cases you can hack things so you can use a newer compiler on a distro that defaults to an older compiler, but it can be pretty difficult to get working in many cases.

CentOS6 and 7: GCC4
Ubuntu/Kubuntu, thru 16.04: GCC5
Arch: GCC6

As you may be aware, CentOS (equivalent to RHEL6 &7) is used on most Linux clusters, because of its stability, and a fair number of users also use it on their desktop for the same reason. Some very large clusters may have enough sysadmins that they have 'modules' which allow for use of other versions of GCC (or perhaps icc), but your typical small local cluster will not.  That means, even if you go to the trouble of compiling EMAN2 from source on these machines, you will still be at a significant performance disadvantage!

Steve Ludtke

unread,
Mar 21, 2017, 9:53:16 AM3/21/17
to EMAN2
New binaries for testing in preparation for the 2.2 release of EMAN2/SPARX/SPHIRE have been posted to the download page (http://ncmi.bcm.edu/ncmi/software/counter_222/software_135). We would encourage people to download them and give them a try. Please report any errors.

If you are currently using a snapshot release, we would encourage you to "mv EMAN2 EMAN2.old" before installing the new binaries, in case there is some problem. The nightly snapshots on the website are currently broken.

The new distributions now use Anaconda Python for all of their library dependencies. This means you get a fully capable install of Anaconda along with your EMAN2 install. We now include tools such as SciPy and Theano with the distribution, and using Anaconda's package manager it is possible to further extend this python environment.

The current Mac binary does not include Pydusa (the new Linux binary does), but we hope to have this issue rectified later today.

The new Linux binaries are compiled with GCC4, so they lack the ~20% performance boost GCC5 can provide, however, they ARE compiled for modern CPUs, meaning these binaries should be ~30% faster than the binaries for all previous EMAN2 releases.

Again, if you have any difficulties with these binaries, please report them ASAP so we can deal with issues before the formal release of 2.2, hopefully sometime in the next week.

cheers

Steve Ludtke

unread,
Jun 2, 2017, 3:39:34 PM6/2/17
to EMAN2
We are pleased to announce the release of EMAN 2.2 (distributed jointly with SPHIRE/SPARX). This new version has been a long time coming and has a very long list of improvements.

For details and downloads, please see:

http://eman2.org

A few of the more interesting improvements since the 2.12 release (full release notes http://eman2.org/Eman22Release) :

Single Particle
- Many refinement changes and new filtering techniques (side-chains look better than Relion or CryoSparc in some cases)
- New bad particle identification strategy, demonstrated to produce better maps
- Local map resolution and filtration
- New automated CTF pipeline  permits very easy(and fast) progressive resolution refinement within a single project folder
- Several new tools for flexibility/dynamics analysis
- New particle picking program (with neural net picker for difficult cases)
- Stochastic Gradient Descent initial model generator (experimental)
- Automatic magnification anisotropy correction tool
- New direct detector movie alignment (experimental)
- New 2-D registration algorithm, faster and more accurate for large box-sizes

Tomography/Subtomogram Averaging
- New subtomogram averaging tools, including up to 20x faster 3-D alignments
- Neural net-based semi-automatic tomogram segmentation (GPU enabled)

General
- New Anaconda Python based distribution with integrated SciPy and other Python libraries
- Source now managed/available through GitHub
- For the first time we have some support for Win10/7 64 bit (not yet well tested, but feel free to try it)

Please see the Google Group for ongoing discussions: http://groups.google.com/group/eman2
or email me directly at any time

Cheers

----------------------------------------------------------------------------
Steven Ludtke, Ph.D.
Professor, Dept of Biochemistry and Mol. Biol.         (www.bcm.edu/biochem)
Co-Director National Center For Macromolecular Imaging        (ncmi.bcm.edu)
Co-Director CIBR Center                          (www.bcm.edu/research/cibr)
Baylor College of Medicine                             
slu...@bcm.edu

Joseph Box

unread,
Jun 5, 2017, 1:14:16 AM6/5/17
to EMAN2
Hey Steve- To get the new release- do we just need to delete our old version of eman2 and then download the appropriate version of the 2.2 release on the eman2 webpage?

Steve Ludtke

unread,
Jun 5, 2017, 10:34:59 AM6/5/17
to em...@googlegroups.com
Hi Tony.  The main reason this release took so long was that we now use a completely different system for installing binaries. Please read the instructions here:


they are not very long, but it is important to read them.


--
--
----------------------------------------------------------------------------------------------
You received this message because you are subscribed to the Google
Groups "EMAN2" group.
To post to this group, send email to em...@googlegroups.com
To unsubscribe from this group, send email to eman2+un...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/eman2

---
You received this message because you are subscribed to the Google Groups "EMAN2" group.
To unsubscribe from this group and stop receiving emails from it, send an email to eman2+un...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Steve Ludtke

unread,
Jun 5, 2017, 10:37:02 AM6/5/17
to em...@googlegroups.com
And why I called you Tony is a mystery to me...   I think I need more sleep.

Mike Strauss

unread,
Jun 6, 2017, 2:01:14 PM6/6/17
to EMAN2
No worries, Clarence.  We all know what it's like.

Steve Ludtke

unread,
Feb 27, 2018, 9:29:04 AM2/27/18
to EMAN2
Announcing the official release of EMAN2/SPARX/SPHIRE 2.21a

You may have noticed that a version called 2.21 was available on the download page for a week or so. We discovered a couple of critical bugs in that release, and decided we needed to post an immediate patch. If you downloaded 2.21, please delete it and replace it with the 2.21a version now available. Note that the downloaded file will still be called 2.21, but after installation e2eversion.py should indicate 2.21a.

Release Notes for EMAN2.21a:
- Improved Neural-network based particle picking. New 3-class method picks even difficult data sets extremely well.
- Initial support for bispectrum based class-averaging and 3-D refinement. ~10-50x faster. 2-D class averaging works very well, 3-D refinement still experimental
- Fix bug causing "missing SNR" problem during refinement in specific situations
- Better support for GPU in Neural Network tomogram segmentation and particle picking
- Direct support for phase plates in CTF correction with adjustable phase slider and autofitting (first version, room for improvement). Issues with using astigmatism and phase plate options together in this version.
- Better GUI display of CTF and Astigmatism
- e2symsearch3d bugs fixed
- Many subtomogram averaging bugs fixed. New pipeline under development.
- Many improvements to e2evalrefine for particle and class-average assessment

------
Release Notes for SPHIRE/SPARX:


Main new components and updates:

- SORT3D – A novel 3D sorting program based on novel concepts of reproducibility. We tested it on several high-resolution projects in our lab and were able to gain valuable and unparalleled insight into data heterogeneity. 
- MERIDIEN – Rapid ML 3D refinement. We have introduced initialization of the program directly from 3D orientation parameters, as opposed to standard mode, which is initiated by an initial template structure. The new mode allows the user to rapidly perform local refinements of data subsets obtained with SORT3D. The two programs, MERIDIEN and SORT3D, thus serve as a perfect tool for exploration of data set heterogeneity. 
- We have streamlined the processing workflow and the GUI interface facilitating users’ interactions with the system. 
- The new release is accompanied by extensive documentation including tutorials explaining the underlying concepts.

For any issues and inquiries, the SPHIRE team is accessible 24/7 via the mailing list. We sincerely hope that the updated SPHIRE 1.0-silver version will meet your expectations and be a helpful advance in high-resolution cryo-EM.

Reply all
Reply to author
Forward
0 new messages