Segment 2D

[Prouteau Manoel]

Dear Carsten,

I am goingslowly throught the workflow of Spring... I am stacked at the 2D segment step.

I run the command:

% segment -f parameterfile.txt

Here is my parameterfile.txt file:

Micrographs                              = *.tif
Image output stack                       = *.hdf
Segment coordinates                      = *.txt
Segment size in Angstrom                 = 1500
Estimated helix width in Angstrom        = 730
Step size of segmentation in Angstrom    = 70
Pixel size in Angstrom                   = 4.5
Invert option                            = True

I get the report.log that I attached to the message. Could you tell me what is the problem?

Thank you very much in advance.

Manoël

[Prouteau Manoel]

Dear Carsten,

I try to keep going on and foun out that I did not created a spring.db file. I did and also used a ney parameterfile.txt to run :

% segment --f parameterfile.txt

Here is the new parameter file:

Micrographs                              = *.tif
Image output stack                       = *.hdf

Spring database option                   = True
spring.db file                           = spring.db
Segment coordinates                      = *boxes.txt

Segment size in Angstrom                 = 1500

Estimated helix width in Angstrom        = 787

Step size of segmentation in Angstrom    = 70

Perturb step option                      = False

Pixel size in Angstrom                   = 4.5

CTF correct option                       = True
CTFFIND or CTFTILT                       = ctftilt
convolve or phase-flip                   = convolve
Binning option                           = False
Binning factor                           = 6
Invert option                            = True
Normalization option                     = True
Row normalization option                 = False
Remove helix ends option                 = False
Rotation option                          = True
MPI option                               = True
Number of CPUs                           = 32
Temporary directory                      = /tmp

Now I get the segment command going further nevertheless it is incapable to create the outfiles *.hdf$
I attach to this message the new report.log of the segment directory.

Thabks in advance for your help,

Manoël

[Carsten Sachse]

Dear Manoel,

The provided report.log says:

ValueError: Specified /home/platform/Documents/TORC1/Superstructure_Spring/spring.db file does not contain micrograph information from CTFFIND. Please re-run MicCtfDetermine.

Have you run MICCTFDETERMINE  and took that spring.db file? In case you don't want to apply CTF correction turn it off in the expert mode.

Best wishes,

Carsten

[Prouteau Manoel]

Dear Carsten,

Thank you very much for your help; I get slowly to understand the work flow of Spring (sorry to be so slow) espacially, thanks to your message, how Spring saves the different files after each step.
I could then solve the problem I had concerning "segment" and the spring.db. I could go then easily through "segmentexam" and "segmentplot". Nevertheless a problem appeared now with:

% segmentalign2d --f parameterfile_segmentalign.txt

The command fails to create a stack.hdf file and the protein_reference_aligned.hdf contains only black scare.

Here is the parameterfile_segmentalign.txt I used:

Image input stack                        = ./segment/stack.hdf
Image output stack                       = stack.hdf
Number of iterations                     = 2
Reference option                         = True
Image reference stack                    = ./segmentref/ref-stack.hdf
Update references                        = True
Aligned average stack                    = protein_reference_aligned.hdf

Pixel size in Angstrom                   = 4.5

Estimated helix width and height in Angstrom = (605, 1500)
Absolute X and Y translation limit in Angstrom = (205, 30)
Internal binning factor                  = 2
Limit in-plane rotation                  = True
Delta in-plane rotation angle            = 10.0
X and Y translation range in Angstrom    = (50, 50)
High-pass filter option                  = False
Low-pass filter option                   = True
High and low-pass filter cutoffs in 1/Angstrom = (0.001, 0.09)
MPI option                               = True
Number of CPUs                           = 8
Temporary directory                      = /tmp

I joined the shell errors and the report.log from the segmentalign2d directory.

I hope that you can help me again. In advance thank you very much.

Have a great week-end.

Best wishes,

Manoël Prouteau, Ph.D.
Scientific Collaborator
Group of Prof. Robbie Loewith
Dept. of Molecular Biology - Sc III
Quai Ernest Ansermet, 30
1211 Geneve 04
Switzerland

[Carsten Sachse]

Dear Manoel,

It looks like there are some image dimension errors. Could you check whether you are using the same image dimension of the reference stack as in your segment stack? This is a pre-requisite to align them.

Best wishes,

Carsten

On Thursday, November 20, 2014 11:40:31 AM UTC+1, Prouteau Manoel wrote:

[Prouteau Manoel]

Dear Carsten,

Indeed the stack.hdf structures were different, I did a mistake in the parameter files in one of the two "segment" commands.
Now I could perform both segmentations and the align2d (parameter files are attached):

% segment --f parameterfile_segment.txt
% segment --f parameterfile_segmentregular.txt
% segmentalign2d --f parameterfile_align.txt

Nevertheless when I run:

% segmentclass --f parameterfile_segmentclass.txt

I get error message (cf parameterfile_segmentclass.txt and report.log). I again dont know what to change; sorry for this step by step work. I hope that I do not bother you too much.

I advance thank you very much for your help.

Have a good end of week-end.

Best,

Manoël

[Carsten Sachse]

Hi Manoel,

In the report file it says:

'segmentalign_mpi did not finish successfully.... Please check log file of segmentalign_mpi for details.'

Segmentclass cycles between k-means clustering and segmentalign2d. Every one of these steps are stored in sub-folders within the segmentclass directory. Double-check the segmentalign2d00 folder and its report.log for details.

Best wishes,

Carsten

[Prouteau Manoel]

Dear Carsten,

Thank you for all your quick answers. I must say that I feel a bit lost. I hope that I am not using wrong parameters in the parameterfiles I sent you.
To simplify, I performed the following command:

% segmentclass --f parameterfile_segmentclass.txt

Using the following parameterfile_segmentclass.txt:

Image input stack                        = ./segment/stack.hdf

Class average stack                      = averages.hdf

Pixel size in Angstrom                   = 4.5

Estimated helix width and height in Angstrom = (610, 820)
Number of classes                        = 5

And I get again the same error message:

IOError: segmentalign2d_mpi did not finish successfully. The output file segmentalign2d01/stack6binnedali.hdf was not found, please check logfile of segmentalign2d_mpi for details.

The report.log file from the segmentalign2d01 directory is attached to this message.

I hope you can help me to go further through the segment2d steps...

In advance thank you for your help.

Manoël

[Carsten Sachse]

Dear Manoel,

I had a look at the results file. It looks like this is a bug I will fix soon. Internally, it cannot deal very well with empty clusters.

How many segments do you have and how many classes did you request?

Decrease your number of classes and your run will become more stable.

Thanks.

Best wishes,

Carsten

On Thursday, November 20, 2014 11:40:31 AM UTC+1, Prouteau Manoel wrote:

[Prouteau Manoel]

Dear Carsten,

Before you answered I check the different subjects in the spring group; I understood that few people had problem with the iterative steps of Spring. Since you adviced them either to play with classes or CPUs numbers I did the same. I have a small number of crops (about a 1000); i play from 3 to 20 classes and from 2 to 32 CPUs none of the combinaisons worked and I got the same error...

Manoël