Difference between revisions of "Viewing tomograms"

From Dynamo
Jump to: navigation, search
Line 45: Line 45:
  
 
== Binned tomograms ==  
 
== Binned tomograms ==  
 +
 +
Binning tomograms produces volumes much easier to fit in memory (1x binning ammounts to 8 times les voxels), and also clearer images. Thus tt is almost always useful to visualize binned versions of the original tomograms, even though the subtomograms for averaging will be cropped from the big, high resolution tomograms.
 +
 +
In all the volume browsers linked to the catalogue the idea is that you shouldn't care about coordinate transformations of annotations produced on different versions of one tomogram. Coordinates of models and particles clicked on binned versions are stored in disk with the correct dimensionality.
 +
 +
=== Binning on-the-fly ===
  
 
=== Prebinned tomograms ===
 
=== Prebinned tomograms ===
  
===
+
=== Explicite volume pairs ===
 +
 
 +
This the most complicated (but most flexible) system. It allows the user to enter in the catalogue two different files to represent one single tomogram. In the column <tt>file</tt> of the [[dcm GUI]], you would introduce the name of a ''depiction volume''. All created annotations will be based on the dimensions of this volume. Now, if you want to actually crop particles from a different volume, you need to enter it in the column <tt> cropping volume </tt>.
 +
 
 +
Once the pair is created, you can put a value of one in the column  <tt>crop elsewhere </tt>. This way, each time you want to crop particles from this tomogram pair, the cropping function will access the tables stored for the ''depiction volume'', scale them according to  <tt>apix crop volume</tt> and <tt>shift crop volume</tt> and then crop the subvolumes from the ''cropping volume''. Note that with this option, you need to specify the <tt>apix</tt> of the ''depiction volume'' itself.
 +
 
 +
An example script is provided in <tt>dpktomo.examples.catalogue.cropHighResPickLowRes</tt>, where you can follow how to define the parameter  <tt>shift crop volume</tt>. This example also shows you how to use this functionality from the command line
 +
 
 +
 
 +
In this case, all model coordinates will be stored with the coordinates of the unbinned depiction tomogram. When the catalogue is requested to crop subtomograms, it will automatically generate cropping tables with coordinates scaled to the size of the ''cropping tomogram''.
 +
 
 +
 
 +
 
 +
This feature is also useful when you want to use tomograms of the same dimensionality but different appearance for picking. For instance, you want to use tomograms reconstructed with SART (typically providing a nice contrast) for depiction and  particle clicking.

Revision as of 13:53, 21 March 2016

Dynamo offers several ways to visualize tomograms. The choice of one or other program depends on varios factors, like "How big is the tomogram?" "Are you going to annotate models on them?".

Tomogram browsers

Most browsers can be invoked directly from the command line or from a GUI linked to a catalogue.

The browsers you are most likely to use are:

  • dtmshow quick, simple browsing outside of the catalogue system
  • dpreview quick inspection of a few slides of catalogued volumes, for sending areas to other browser
  • dtmview annotation of isolated particles and membranes inside the catalogue system
  • dtmslice annotation of filaments and membranes inside the catalogue system

dtmshow

The simplest visualization system is the program dtmshow, callable from the command line and linked by many Dynamo GUIs. It's a lightweight slice-based browser. The user can choose to use it by preloading of the full tomogram or by accessing the disk for each depicted slice ("on-the-fly" modus). dtmshowDoes not allow to make complex annotations, or to register them inside the Catalogue via the model pool

dpreview

A screenshot on dpreview

dpreview goes a step further than dtmshow. It is connected with the Catalogue system, in fact 'open with dpreview' is one of the main options that you get when selecting a catalogued tomogram in the dcm GUI.

In dpreviewslices are always accessed "on the fly", so that browsing can be slow (especially if trying to depict slices orthogonal to the x- or y- directions, which demand multiple I/O operations to gather a slice whose pixels are scattered in the disk). The main function of the dpreview is to provide a "preview" of the disk contents of a tomogram, letting the user to specify and area of the tomogram and a binning level. This fragment will be read into memory and passed to dtmview or dtslice


dtmview

Is normally invoked by the dcm GUI or by dpreview . It is oriented to show concurrently slices representing orthogonal cuts inside a volume with a high level of control of the number of depicted slices.

The main utility of dtmview is for clicking sets of isolated particles.


dtslicer

Is normally invoked by the dcm GUI or by dpreview . It depicts in 3d a slide or arbitrary orientation that can be moved freely inside the tomogram. This is a suitable browser for filament geometries. It is also suitable to locate the extent of membranes a send them over to the Sliding Montage browser for annotation.


Sliding montage

Sliding montage generated fromdtslice

A sliding montage presents simultaneously a selection of slides of the tomogram. The user can choose to distance between consecutively depicted slices, and navigate inside the tomogram.

Sliding montages of areas inside a tomogram are specially useful for modelling membranes. Sliding montages are connected with the model pool and provide semiautomatic tools to track points manually picked in one level on the following levels.

Binned tomograms

Binning tomograms produces volumes much easier to fit in memory (1x binning ammounts to 8 times les voxels), and also clearer images. Thus tt is almost always useful to visualize binned versions of the original tomograms, even though the subtomograms for averaging will be cropped from the big, high resolution tomograms.

In all the volume browsers linked to the catalogue the idea is that you shouldn't care about coordinate transformations of annotations produced on different versions of one tomogram. Coordinates of models and particles clicked on binned versions are stored in disk with the correct dimensionality.

Binning on-the-fly

Prebinned tomograms

Explicite volume pairs

This the most complicated (but most flexible) system. It allows the user to enter in the catalogue two different files to represent one single tomogram. In the column file of the dcm GUI, you would introduce the name of a depiction volume. All created annotations will be based on the dimensions of this volume. Now, if you want to actually crop particles from a different volume, you need to enter it in the column cropping volume .

Once the pair is created, you can put a value of one in the column crop elsewhere . This way, each time you want to crop particles from this tomogram pair, the cropping function will access the tables stored for the depiction volume, scale them according to apix crop volume and shift crop volume and then crop the subvolumes from the cropping volume. Note that with this option, you need to specify the apix of the depiction volume itself.

An example script is provided in dpktomo.examples.catalogue.cropHighResPickLowRes, where you can follow how to define the parameter shift crop volume. This example also shows you how to use this functionality from the command line


In this case, all model coordinates will be stored with the coordinates of the unbinned depiction tomogram. When the catalogue is requested to crop subtomograms, it will automatically generate cropping tables with coordinates scaled to the size of the cropping tomogram.


This feature is also useful when you want to use tomograms of the same dimensionality but different appearance for picking. For instance, you want to use tomograms reconstructed with SART (typically providing a nice contrast) for depiction and particle clicking.