A question about dynamo2warp

[yuqi zhang]

When I did the tutorial of teamtomo.org, I met a problem about dynamo2warp. After finishing the subset_table_based_on_neighbours, I could not extract subtomo particles with dynamo2warp. The annoying point was that the relion_reconstruct could not give me a hexagonal lattice map. There was only noise. 

I have tried specified different parameter of subtomo particles extract in warp. It did not work. 

So I suspected whether the process of removing the trash particles had error. Then, I tried to align the particles with the 'result_10Apx_nodup_neighbourcleaning.tbl', which has no trash particles, in the tutorial in dynamo. However, the alignment result was correct.

I think maybe the dynamo2warp script went wrong. Can someone teach me something?

[Alister Burt]

Hi Yuqi,

If the reconstruction in Dynamo is working (gives a good average) and the reconstruction of particles in Warp isn’t working (running relion_reconstruct on the particles gives a bad average) then this means the particles you reconstruct in Warp are incorrect, something in the metadata (positions/orientations) is wrong.

This could be for a few different reasons

1. Your tomogram dimensions in Warp changed, your particle centers from Dynamo refer to different tomogram dimensions and are now incorrect

2. You input some parameters wrong when reconstructing particles from the STAR file (pixel size for STAR file)

3. dynamo2warp is broken

dynamo2warp is passing all tests and others have successfully used it to work between Dynamo and Warp so I think it’s more likely 1 or 2. If you check points 1 and 2 carefully then provide more information about the procedures followed, settings in warp, tomogram dimensions etc then we can investigate the possibility of 3.

Good luck,

Alister

--
You received this message because you are subscribed to the Google Groups "Warp" group.
To unsubscribe from this group and stop receiving emails from it, send an email to warp-em+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/warp-em/7d695c38-d9cf-4b19-b7ce-3cef2c153aaan%40googlegroups.com.

[yuqi zhang]

Thank you for your so fast reply! I didn't know about the first reason in your statement. What is the meaning of tomogram dimensions? The other two are impossible in my case. I have ever checked these two points.

[Pranav Shah]

Hi, What is the output of running the header command (from imod) on of
your tomograms? For example in my case i get the following values -

pss-MacBook-Pro:Position_05 ps$ header Position_05_rec_deconv.mrc

RO image file on unit 1 : Position_05_rec_deconv.mrc Size= 1359379 K

Number of columns, rows, sections ..... 1366 1366 373
Map mode .............................. 1 (16-bit integer)
Start cols, rows, sects, grid x,y,z ... 0 0 0 1366 1366 373
Pixel spacing (Angstroms).............. 1.000 1.000 1.000
Cell angles ........................... 0.000 0.000 0.000
Fast, medium, slow axes ............... X Y Z
Origin on x,y,z ....................... 0.000 0.000 0.000
Minimum density ....................... -5834.0
Maximum density ....................... 3292.0
Mean density .......................... 506.22
tilt angles (original,current) ........ 0.0 0.0 0.0 0.0 0.0 0.0
Space group,# extra bytes,idtype,lens . 0 0 0 0

0 Titles :

Secondly, what value did you enter in the unbinned tomogram dimensions
dialog box (see image)?
Finally, please put the effort in giving us as much information as you
can about your experimental parameters, as it becomes tedious to
answer questions with minimal information. Sometimes if something is
unclear it is better to read the docs yourself
(http://www.warpem.com/warp/?page_id=378) rather than pestering the
group...

Best,
Pranav
--
Pranav Shah
Postdoctoral Research Fellow.

Division of Structural Biology,
Wellcome Trust Centre for Human Genetics,
University of Oxford,
Roosevelt Drive, Oxford OX3 7BN,
UK

> To view this discussion on the web visit https://groups.google.com/d/msgid/warp-em/7ae68934-0aff-4025-857e-80281c99e939n%40googlegroups.com.

[yuqi zhang]

Thank you very much.

The header command:

$ header TS_01.mrc_10.00Apx.mrc

 RO image file on unit   1 : TS_01.mrc_10.00Apx.mrc     Size=     460182 K

 Number of columns, rows, sections .....     540     540     404

 Map mode ..............................    2   (32-bit real)

 Start cols, rows, sects, grid x,y,z ...    0     0     0     540    540    404

 Pixel spacing (Angstroms)..............   10.00      10.00      10.00

 Cell angles ...........................   90.000   90.000   90.000

 Fast, medium, slow axes ...............    X    Y    Z

 Origin on x,y,z .......................    0.000       0.000       0.000

 Minimum density ....................... -0.36994

 Maximum density .......................  0.47631

 Mean density ..........................   0.0000

 tilt angles (original,current) ........   0.0   0.0   0.0   0.0   0.0   0.0

 Space group,# extra bytes,idtype,lens .        0        0        0        0

     0 Titles :

The warp GUI:

[yuqi zhang]

And I think maybe the version of relion affects the results. I suspect the parameter:

_rlnPixelSize #13

_rlnVoltage #14

_rlnSphericalAberration #15

When I used relion 3.0.6, relion_reconstruct can be better.

The head of subtomograms.star was like:

data_

loop_

_rlnCoordinateX #1

_rlnCoordinateY #2

_rlnCoordinateZ #3

_rlnAngleRot #4

_rlnAngleTilt #5

_rlnAnglePsi #6

_rlnMicrographName #7

_rlnMagnification #8

_rlnDetectorPixelSize #9

_rlnCtfMaxResolution #10

_rlnImageName #11

_rlnCtfImage #12

   124.907   808.545  315.739    20.590000  171.390000  -158.380000  TS_03.mrc.tomostar  10000.0  5.00000  4.9  subtomo/TS_03.mrc/TS_03.mrc_0000000_5.00A.mrc  subtomo/TS_03.mrc/TS_03.mrc_0000000_ctf_5.00A.mrc

When I used relion 3.1.0, relion_reconstruct was nothing.

The head of subtomograms.star was like:

data_

loop_

_rlnCoordinateX #1

_rlnCoordinateY #2

_rlnCoordinateZ #3

_rlnAngleRot #4

_rlnAngleTilt #5

_rlnAnglePsi #6

_rlnMicrographName #7

_rlnMagnification #8

_rlnDetectorPixelSize #9

_rlnCtfMaxResolution #10

_rlnImageName #11

_rlnCtfImage #12

_rlnPixelSize #13

_rlnVoltage #14

_rlnSphericalAberration #15

   124.907   808.545  315.739    20.590000  171.390000  -158.380000  TS_03.mrc.tomostar  10000.0  5.00000  4.9  subtomo/TS_03.mrc/TS_03.mrc_0000000_5.00A.mrc  subtomo/TS_03.mrc/TS_03.mrc_0000000_ctf_5.00A.mrc 5.0000 300.0000 2.70000

And the command line I used was the same:

mpirun -np 16 `which relion_reconstruct_mpi` --i subtomograms.star --o rec.mrc --3d_rot --ctf --sym C6

[yuqi zhang]

I should show you the screenshot of map in 3.1.0 & 3.0.6

[Alister Burt]

Hi Yuqi, 

The difference here is not due to you using different versions of RELION -> the first star file is straight after particle reconstruction in warp, the second is after an alignment in RELION.

I understand that when things don’t work for you it can be frustrating. You need to be methodical, reconstruct the particles in Dynamo -> use dynamo2warp -> reconstruct the particles in RELION then compare the results. Do not perform any refinements in between these steps. If at that point the reconstructions differ, then you can take steps to figure out why

Cheers,

Alister

On 11 Jun 2021, at 05:27, yuqi zhang <8245...@qq.com> wrote:

And I think maybe the version of relion affects the results. I suspect the parameter:

To view this discussion on the web visit https://groups.google.com/d/msgid/warp-em/b4229071-7ed0-46fa-bde9-c1e96ca103a7n%40googlegroups.com.

[Alister Burt]

Amending this

The difference here is not due to you using different versions of RELION -> the first star file is straight after particle reconstruction in warp, the second is after an alignment in RELION.

The second star file appears to have been modified by relion at some point (contains extra information compared to the dynamo2warp output). It doesn’t contain shifts so is probably not from a refinement

[yuqi zhang]

Was the relion_reconstruct result in the screenshot (relion3.0.6) good enough? I think it was also too noisy. 

To avoid any wrong thing, I think I should do the tutorial ab initial.