Soon relion5 may improve some of this
October 6, 2023 at 9:50 PM
Soon relion5 may improve some of this
We're pleased to announce that #RELION5 is now available for beta testing! 🥳🥳🥳 Download it for free from github.com/3dem/relion/...
#OpenSoftwareAcceleratesScience
#OpenSoftwareAcceleratesScience
October 26, 2023 at 4:22 PM
We're pleased to announce that #RELION5 is now available for beta testing! 🥳🥳🥳 Download it for free from github.com/3dem/relion/...
#OpenSoftwareAcceleratesScience
#OpenSoftwareAcceleratesScience
Warptools can load alignments and then export to relion5 format!
November 11, 2024 at 4:21 AM
Warptools can load alignments and then export to relion5 format!
Finally got some time to properly test #RELION5 complete pipeline, starting from raw frames to STA. 22 tilt-series (from EMPIAR-11830), picking done with pytom, 14600 final particles, FSC 4.1-ish on the LSU (Nyquist bin1 3.8). Ofc it's a ribosome 😅 but now I can try with smthng more chlengin
#CryoET
#CryoET
November 10, 2024 at 7:11 PM
Now on @biorxivpreprint.bsky.social: Dari Kimanius' paper on #Blush regularisation, a machine-learning based prior in #RELION5 trained on EMDB #cryoEM half-maps, which he uses to solve a #cryoEM data set of a 40 kDa protein:RNA complex that was previously intractable!
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...
October 24, 2023 at 12:40 PM
Now on @biorxivpreprint.bsky.social: Dari Kimanius' paper on #Blush regularisation, a machine-learning based prior in #RELION5 trained on EMDB #cryoEM half-maps, which he uses to solve a #cryoEM data set of a 40 kDa protein:RNA complex that was previously intractable!
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...
We decided to summarize our (the @cellarchlab.com group) most common workflows! The Tutorial guides you through tomogram pre- and postprocessing using either #RELION5, #Scipion, or #AreTomo3. With the Chlamy dataset (www.ebi.ac.uk/empiar/EMPIA...) as a possible resource for all those steps!
2/🧵
2/🧵
May 6, 2025 at 4:27 PM
We decided to summarize our (the @cellarchlab.com group) most common workflows! The Tutorial guides you through tomogram pre- and postprocessing using either #RELION5, #Scipion, or #AreTomo3. With the Chlamy dataset (www.ebi.ac.uk/empiar/EMPIA...) as a possible resource for all those steps!
2/🧵
2/🧵
Check out 'the website' by Phil&Flo - a tomo processing pipeline/tutorial from frame to structure. For easy implementation of the Aretomo3 to RELION5 pipeline, or your students first dataset, get processing!
May 6, 2025 at 10:10 PM
Check out 'the website' by Phil&Flo - a tomo processing pipeline/tutorial from frame to structure. For easy implementation of the Aretomo3 to RELION5 pipeline, or your students first dataset, get processing!
And our new #DynaMight paper is out on @biorxivpreprint.bsky.social! 🥳 This is #RELION5's answer to modelling molecular flexibility by the amazing Johannes Schwab. He also shows that model bias can be nasty when modelling molecular flexibility with many parameters.
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...
October 19, 2023 at 3:32 PM
And our new #DynaMight paper is out on @biorxivpreprint.bsky.social! 🥳 This is #RELION5's answer to modelling molecular flexibility by the amazing Johannes Schwab. He also shows that model bias can be nasty when modelling molecular flexibility with many parameters.
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...
Using the amazing #pytom-match-pick we guide you through particle picking. From the template matching results we show you how to perform STA in #RELION5 pushing Ribosomes to their absolute limit (Nyquist)!
3 /🧵
3 /🧵
May 6, 2025 at 4:27 PM
Using the amazing #pytom-match-pick we guide you through particle picking. From the template matching results we show you how to perform STA in #RELION5 pushing Ribosomes to their absolute limit (Nyquist)!
3 /🧵
3 /🧵
Next up in our preparation for #RELION5 will be a
@biorxivpreprint.bsky.social to describe #DynaMight: our solution for ’exploring protein *Dyna*mics that *Might* improve your map’ by the amazing Johannes Schwab. #staytuned #cryoEM
@biorxivpreprint.bsky.social to describe #DynaMight: our solution for ’exploring protein *Dyna*mics that *Might* improve your map’ by the amazing Johannes Schwab. #staytuned #cryoEM
October 18, 2023 at 8:39 AM
Next up in our preparation for #RELION5 will be a
@biorxivpreprint.bsky.social to describe #DynaMight: our solution for ’exploring protein *Dyna*mics that *Might* improve your map’ by the amazing Johannes Schwab. #staytuned #cryoEM
@biorxivpreprint.bsky.social to describe #DynaMight: our solution for ’exploring protein *Dyna*mics that *Might* improve your map’ by the amazing Johannes Schwab. #staytuned #cryoEM
The #OpenSoftwareAcceleratesScience code can already be downloaded from the link below. However, it may be worth waiting another week or so, when we release #RELION5 for easier installation and full integration in its GUI.
October 19, 2023 at 3:33 PM
The #OpenSoftwareAcceleratesScience code can already be downloaded from the link below. However, it may be worth waiting another week or so, when we release #RELION5 for easier installation and full integration in its GUI.
I also feel it’s easier to have misalignment 3d refinement in relion5 than relion4 when dealing with particles restrained by other densities like membrane attached proteins. Although I have not tried 3d particles yet
December 20, 2025 at 1:17 AM
I also feel it’s easier to have misalignment 3d refinement in relion5 than relion4 when dealing with particles restrained by other densities like membrane attached proteins. Although I have not tried 3d particles yet
Yes indeed! This is the only time i went "out" of RELION for this test. pytom-match-pick has been working very well for my targets, and directly outputs RELION5 compatible coordinate files.
November 10, 2024 at 8:32 PM
Yes indeed! This is the only time i went "out" of RELION for this test. pytom-match-pick has been working very well for my targets, and directly outputs RELION5 compatible coordinate files.
Just pushed a beta release of geopickr with star (relion5 and warp) and dynamo tbl output. I don't have a pressing problem to test these out on, but if you have a workflow that would benefit from one of these, want to test it out and give me feedback? github.com/baradlab/Geo...
Release Geopickr 1.1.0b1 (beta) · baradlab/Geopickr
Beta pre-release adding multi-format particle export, plus a Place Object fix.
New: export to Dynamo & RELION (issues #1, #2 — pending live-import testing)
Dynamo .tbl (+ optional .vll volume-list...
github.com
June 19, 2026 at 9:36 PM
Just pushed a beta release of geopickr with star (relion5 and warp) and dynamo tbl output. I don't have a pressing problem to test these out on, but if you have a workflow that would benefit from one of these, want to test it out and give me feedback? github.com/baradlab/Geo...
So, for my test case the pseudosubtomos (3D option in Relion5) already provided a better map than the 2D stacks but still not as detailed and complete as warp 3D subtomos in Relion3.
December 12, 2025 at 3:38 PM
So, for my test case the pseudosubtomos (3D option in Relion5) already provided a better map than the 2D stacks but still not as detailed and complete as warp 3D subtomos in Relion3.
Do you mean 3D as in Warp-exported 3D subvolumes+CTF (i.e. Relion3 format) or pseudo-subtomograms as in Relion4 --tomo? AFAIK the latter is supported in Relion5 but not the former, as @jpkreysing.bsky.social said
December 12, 2025 at 3:14 PM
Do you mean 3D as in Warp-exported 3D subvolumes+CTF (i.e. Relion3 format) or pseudo-subtomograms as in Relion4 --tomo? AFAIK the latter is supported in Relion5 but not the former, as @jpkreysing.bsky.social said
Blush is a denoising neural network that is trained on many EMDB #cryoEM half-maps using a noise2noise framework. It replaces the Fourier-based (smoothness) prior inside the iterative RELION refinement. That @biorxivpreprint.bsky.social will be next, as also Blush will be part of #RELION5.
October 18, 2023 at 1:56 PM
Blush is a denoising neural network that is trained on many EMDB #cryoEM half-maps using a noise2noise framework. It replaces the Fourier-based (smoothness) prior inside the iterative RELION refinement. That @biorxivpreprint.bsky.social will be next, as also Blush will be part of #RELION5.
No worries. You could try 3D classification to select a more homogeneous subset of particles. But, as the spike is rather floppy, this may be difficult. Soon, we will release Dynamight, a new tool in #RELION5 by Johannes Schwab that models protein "Dyna"mics, which "might" help. 😜
September 27, 2023 at 8:31 AM
No worries. You could try 3D classification to select a more homogeneous subset of particles. But, as the spike is rather floppy, this may be difficult. Soon, we will release Dynamight, a new tool in #RELION5 by Johannes Schwab that models protein "Dyna"mics, which "might" help. 😜
Confirmed by gold-standard FSC of course :)
pytom-match-pick does not use projections unlike RELION5 and I suppose there is also quite the difference from the previous pseudo-subtomos
pytom-match-pick does not use projections unlike RELION5 and I suppose there is also quite the difference from the previous pseudo-subtomos
March 14, 2026 at 1:09 PM
Confirmed by gold-standard FSC of course :)
pytom-match-pick does not use projections unlike RELION5 and I suppose there is also quite the difference from the previous pseudo-subtomos
pytom-match-pick does not use projections unlike RELION5 and I suppose there is also quite the difference from the previous pseudo-subtomos
I currently still use 3D warp subtomos in Relion3 because either version of 2D input data (pseudosubtomos or 2D stacks) give considerably worse results in Relion5 (tested for small particles).
December 12, 2025 at 11:43 AM
I currently still use 3D warp subtomos in Relion3 because either version of 2D input data (pseudosubtomos or 2D stacks) give considerably worse results in Relion5 (tested for small particles).
Can you use 3D subtomos in Relion5? I recall a message from Sjors at some point that this is no longer supported which matches my experience (from maybe 6 months ago).
December 12, 2025 at 11:44 AM
Can you use 3D subtomos in Relion5? I recall a message from Sjors at some point that this is no longer supported which matches my experience (from maybe 6 months ago).
This was the last preprint before our #RELION5 beta release, which I will hopefully get to by tomorrow afternoon or on Thursday. #staytuned
October 24, 2023 at 12:45 PM
This was the last preprint before our #RELION5 beta release, which I will hopefully get to by tomorrow afternoon or on Thursday. #staytuned
To get a high-resolution reconstruction, we tried a lot of things, but the map quality was never all that great. Until BLUSH regularisation in RELION5 came along and provided us with an isotropic map into which we were able to properly refine a model that also agreed with AlphaFold predictions.
February 17, 2024 at 3:44 PM
To get a high-resolution reconstruction, we tried a lot of things, but the map quality was never all that great. Until BLUSH regularisation in RELION5 came along and provided us with an isotropic map into which we were able to properly refine a model that also agreed with AlphaFold predictions.