Note: Denoising micrographs before preprocessing and before binning will make particle picking easier. Training on denoised micrographs may not work any better than the originals. It is recommended that you denoise micrographs for manual picking/inspection by using the Denoise tab on the right.
The radius of the displayed pick is 5 pixels in 1x zoom images
Downsample further if particle radius in pixels is > 35 (resnet8), 15 (conv31), 31 (conv63), 63 (conv127)
Pick 100-1000 particles (more is better)
Pick all representative views
Pick from several micrographs
Incomplete picking is fine!
Available only on image focus
Move selected pick by 1 pixel (+ Shift for 10 pixels)
Select all picks
Copy selected picks
Paste selected picks
Delete selected picks
+-= / Ctrl + Mouse Wheel
Zoom in/out (cursor over image)
Toggle pick number/label
Toggle pick boundary
Toggle cursor coordinates
Upload images from disk
Move to next/previous image
Toggle Topaz score (after training)
Toggle pick annotation editor (Ctrl to toggle on image editor)
Home / h
Jump to first image
End / e
Jump to last image
PageUp / u
Jump several images
PageDown / d
Jump several images
Cancel ongoing task
Pick AttributesImage Attributes
Toggle Annotation Editor
Selected0of0images in current group, show
If all images in your project are saved in a single folder, set the default path to the location of this folder. The VIA application will load images from this folder by default. Note: a default path of "./" indicates that the folder containing the topaz.html application file also contains the images in this project. For example: /datasets/VOC2012/JPEGImages/ or C:\Documents\data\(note the trailing / and \)
Search Path List
If you define multiple paths, all these folders will be searched to find images in this project. We do not recommend this approach as it is computationally expensive to search for images in multiple folders.
Particle Pick Label
By default, each pick in an image is labelled using the region-id. Here, you can select a more descriptive labelling of regions.
By default, each pick is drawn using a single color. Using this setting, you can assign a unique color to picks grouped according to an attribute such as Topaz score.
Particle Pick Label Font
Font size and font family for showing pick labels.
Preload Buffer Size
Images are preloaded in buffer to allow smoother navigation of next/prev images. A large buffer size may slow down the overall browser performance. To disable preloading, set buffer size to 0.
On-image Annotation Editor
When a single particle pick is selected, the on-image annotation editor is activated which the user may use to update annotations of this pick. By default, this on-image annotation editor is placed near the selected pick.
File Not Found
We recommend that you update the default path in project settings to the folder which contains this image.
A temporary fix is to use browser's file selector to manually locate and add this file. We do not recommend this approach because it requires you to repeat this process every time your load this project in the VIA application.
Denoise MRCs (recommended): Use Topaz-Denoise to denoise full-size raw micrographs.
Denoising using the pre-trained models works best on K2, K3, Falcon II, and Falcon III micrographs. You may wish to make your own denoising model if the pre-trained models are not optimal.
Preprocess MRCs: Use Topaz to downsample, normalize, and convert denoised micrographs to PNG (for use in Topaz picking GUI) and raw micrographs to MRC (for use with Topaz training).
Your particle must have a diameter (longest dimension) after downsampling of: 70 pixels or less for resnet8; 30 pixels or less for conv31; 62 pixels or less for conv63; 126 pixels or less for conv127.
Topaz normalization uses image statistics for all input micrographs. A Gaussian mixture model is used to effectively make the ice region of each micrograph visible, even if there is gold in the image.
Pick Particles: Add images then pick particles:
To add images:
Click Add Images or Add Image Paths under 'Project' to the left
Note: This Topaz GUI accepts PNG, JPEG, and BMP files.
To pick particles:
Left-click on the center of particles for several micrographs.
To zoom in/out, either press Ctrl and scroll your mouse wheel on an image or press + or - (equal = will reset the zoom).
To delete a misplaced point, click the point again and press d.
To adjust a point's location, click the point again and either use the arrow keys on your keyboard or click-and-drag to move it.
Note: Topaz works best if you identify 500+ particles representing all orientations of the particle and if you pick across several representative micrographs. More training particles and views means more accurate Topaz-trained picks.
Note: The radius of the displayed pick is 5 pixels in un-zoomed images.
Train a model: Use Topaz to train a neural network model to pick your particles.
Training often works best when the training radius is 3 pixels or less.
Several parameters are optimizable by cross-validation.
Export Particles: To export particles, click Particle Picks → Export Picks in the top menu. These picks may then be used in Topaz training.
Press L on your keyboard to show the particle number, then the Up ↑ key will toggle the Topaz particle scores. Zoom-in on the image with + to inspect.
Note: It is good practice to save your particles/project regularly. Do this by clicking Particle Picks → Export Picks or Project → Save in the top menu.
Note: Parameters are categorized by those that need to be modified, might need to be modified, are optimizable, and are rarely modified. Training parameters that are optimizable may be optimized using cross-validation.
If you wish to work outside this GUI, follow the GitHub tutorials.
Topaz is a pipeline for particle detection in cryo-electron microscopy images using convolutional neural networks trained from positive and unlabeled examples. The original research can be found here. Topaz is released under the GPLv3 license.
Topaz GUI uses VGG Image Annotator (VIA), which is an image annotation tool that can be used to define regions in an image and create textual descriptions of those regions. VIA is an open source project developed at the Visual Geometry Group and released under the BSD-2 clause license.
Here is a list of some salient features of the Topaz GUI version of VIA:
can be used off-line (full application in a single HTML file that is < 0.5MB)
only requires a modern web browser (tested on Chrome, Firefox, Edge, and Safari; Note: Some features look/work differently in different browsers; Chrome is recommended)
Copyright (c) 2016-2019, Abhishek Dutta, Visual Geometry Group, Oxford University and VIA Contributors.
All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
Redistributions of source code must retain the above copyright notice, this
list of conditions and the following disclaimer.
Redistributions in binary form must reproduce the above copyright notice,
this list of conditions and the following disclaimer in the documentation
and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
Pick or Analyze Particles
Pick particles for training networks or analyze Topaz picks. Follow the steps below:
Particle picking for training
1) To start picking, select images or add images from paths.2) Pick 100-1,000 particles (more is better) representing all orientations.
• Pick across several representative images.
• Not all particles need to be picked from each image.
3) Export your particle picks and proceed to Topaz training.
• Save your particles and project regularly. You may loadthem later to continue.
• For further help, see the Getting Started page.
Analyzing Topaz picks
• Import your Topaz particle picks as a CSV file and review the picks.
◦ Press L on your keyboard to show the particle number, then the Up ↑ key will toggle the Topaz particle scores. Zoom-in on the image with + to inspect.
• Load your micrographs and particles in .star format into Relion. Picks may be colored based on Topaz score by using the AutopickFigureOfMerit column in the .star file.
Topaz Training Command Generator
Create a command for training a model using particle picks.