Point region shape n x

Welcome!

This GUI will allow you to pick training particles and make Topaz commands. Here is the general workflow:
→ → →
Normalize &
downsample images		 Pick training
particles, or
analyze picks		 Train a
neural network		 Extract particle
coordinates


Citation: Bepler, T., Morin, A., Brasch, J., Shapiro, L., Noble, A.J., Berger, B. (2018). Positive-unlabeled convolutional neural networks for particle picking in cryo-electron micrographs. arXiv.

      Topaz Image Pre-processing Command Generator

Create a command for pre-processing images (downsampling and normalization) for use in this Topaz GUI and in Topaz training.



Parameters:            Must modify Might modify Rarely modify


Input micrographs
Output folder
Number of CPUs
Scale factor

Advanced options
Output format
Pixel sampling
Number of iterations
Random seed


Command:

topaz preprocess /path/to/input/images/*.mrc --scale 8 --num-workers -1 --format mrc,png --pixel-sampling 25 --niters 200 --seed 1 --verbose --destdir /path/to/output/preprocessed/images/
Pick Annotations Image Annotations × ↑ ↓ + −
  • Home
  • Project
    • Load
    • Save
    • Settings
    • Add local images
    • Add images from absolute paths
    • Add images from URL
    • Add urls or paths from text file
  • Particle Picks
    • Export Picks (as csv)
    • Import Picks (from csv)
    • Preview Picks
    • Download as Image
  • View
    • Toggle image grid view
    • Toggle left sidebar
    • Toggle image filename list
    • Toggle attributes editor
    • Toggle annotation editor (Space)
    • Show/hide pick boundaries (b)
    • Show/hide pick labels (l)
  • Help
Open Project Save Project Update Project Settings Import Picks (as csv) Export Picks (as csv) Switch to Image Grid View Previous Image Toggle Image List Next Image Zoom In Zoom Out Select All Picks Copy Picks Paste Picks Paste Pick in Multiple Images Undo Picks Pasted in Multiple Images Delete Pick
Sorry, your browser does not support HTML5 Canvas functionality which is required for this application.
Group by 
Selected 0 of 0 images in current group, show

Settings

Project Name
Default Path
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 region in an image is labelled using the region-id. Here, you can select a more descriptive labelling of regions.
    Particle Pick Label Font
    Font size and font family for showing region 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 region.

    File Not Found

    Filename:

    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.

    1. Preprocess MRCs: Use Topaz to downsample, normalize, and convert micrographs to PNG (for use in Topaz GUI) and MRC (for use with Topaz training).
      • Your particle must have a diameter (longest dimension) of 30 pixels or less after downsampling
      • 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.
    2. 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 images. More training particles and views means more accurate Topaz-trained picks.
        • Note: The radius of the displayed pick is 5 pixels.
    3. Train a model: Use Topaz to train a neural network model to pick your particles.
      • Several parameters are optimizable by cross-validation.
    4. Export Particles: To export particles, click Particle Picks → Export Picks in the top menu. These picks may then be used in Topaz training.
    5. Analyze Topaz Picks: Load Topaz picks by clicking Particle Picks → Import Picks.

      6. 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.
    Loading ...
        
    Topaz Version 0.1.0VIA Version 2.0.5

    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:

    • based solely on HTML, CSS and Javascript
    • 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, and Safari)
    • import/export particle picks in csv file format

    For more Topaz details, visit https://github.com/tbepler/topaz/.

    For more VIA details, visit http://www.robots.ox.ac.uk/~vgg/software/via/.

     

    VIA is Copyright © 2016-2018, Abhishek Dutta,Visual Geometry Group, Oxford University and VIA Contributors.

    Copyright (c) 2016-2018, 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 absolute path.
      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 load them 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.

      Or

      • 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 network using particle picks.



    Parameters:            Must modify Might modify Optimizable Rarely modify


    Training images folder
    Training particles
    Output
    Particle radius
    Autoencoder
    Number of epochs
    GPU/CPU Device
    Number of CPUs
    CNN model
    Method
    Num particles/image
    K-fold

    Advanced options
    Fold
    Image extension
    Units
    Dropout
    Batch norm
    Unit scaling
    Scaled num of units
    L2
    Learning rate
    Minibatch size
    Minibatch balance
    Epoch size
    Test batch size


    Command:

    topaz train --train-images /path/to/preprocessed/images/ --train-targets /path/to/training_particles.csv --k-fold 5 --fold 0 --radius 3 --model resnet8 --image-ext .mrc --units 32 --dropout 0.0 --bn on --unit-scaling 2 --ngf 32 --method GE-binomial --autoencoder 0 --num-particles 300 --l2 0 --learning-rate 0.0002 --minibatch-size 256 --minibatch-balance 0.0625 --epoch-size 5000 --num-epochs 10 --num-workers -1 --test-batch-size 1 --device 0 --save-prefix /output/path/model --output /output/path/results.txt

       Topaz Picks Extraction Command Generator

    Create commands for extracting and converting particle coordinates.



    Parameters:            Must modify Might modify Rarely modify


    Input micrographs
    Model
    Output filenames
    Particle radius
    Upscale picks
    Number of CPUs
    GPU/CPU Device

    Advanced options
    Threshold
    Minimum radius
    Maximum radius
    Step radius


    Commands:

    topaz extract /path/to/preprocessed/images/*.mrc --model /path/to/model_epoch##.sav --radius 8 --threshold 0.5 --up-scale 1 --min-radius 5 --max-radius 100 --step-radius 5 --num-workers -1 --device 0 --output /path/to/extracted/particles.txt

    topaz convert /path/to/extracted/particles.txt --verbose 1 --output /path/to/extracted/particles.star

       More Tools

    Create commands for converting/scaling particle coordinates, converting images, and scaling images.


    Convert/Scale coordinates

    Parameters:            Must modify Might modify Rarely modify

    Convert images

    Parameters:            Must modify Might modify Rarely modify

    Scale images

    Parameters:            Must modify Might modify Rarely modify