Skip to content

Loading data

Jump to bottom
opambos edited this page Apr 21, 2026 · 2 revisions

Loading a new dataset

DeepTRACE can accept localisation or tracking data from the following pipelines:

Picasso-formatted data

We recommend the use of the software Picasso (https://github.com/jungmannlab/picasso) from the Jungmann lab. When using this input method, DeepTRACE will use its own internal nearest-neighbour linker to connect localisations (grouped by segmented cell boundaries) into tracks.

Preparation of your data

  • Open Picasso and perform localisation using LQ, Gaussian fitting (produces .hdf5 file).
  • Segment cells in your image using the software MicrobeTracker (produces .mat file).

Procedure

  1. Click File > Load new localisation data to launch the Select Source Files dialog (screenshot below).
  2. Select the reference image, MicrobeTracker-formatted cell boundaries, Picasso localisation files, and the fluorescence recording for a single field of view. For multiple recordings see the note below.
  3. Enter an inter-frame time in seconds and press enter; this defines the time between the start of each frame, equal to the exposure time plus any readout time.
  4. Click OK to continue; you will now be presented with the Data Import GUI (screenshot below).
  5. Use the arrows to remove any offset between the localisations and cell boundaries; for some pipelines the definition of y-axis can be inverted, if so click Flip image and cell outlines.
  6. Once aligned click Set static offset and exit.
  7. To perform nearest-neighbour linking you will be asked to provide a memory parameter and a linking distance.
  8. You will now be presented with Data Preparation GUI (screenshot below); detailed descriptions for each variable can be found by hovering over each input. Click OK.
  9. When prompted, provide a dataset name that will allow to you identify the dataset.
  10. Click OK to continue to the Feature Engineering menu, which is described in the next page of the wiki.

Note about multiple recordings: If you provide multiple recordings DeepTRACE will display a separate GUI prompting the user to sort recordings chronologically; it is important that the order matches exactly the localisation files. The Select Source Files dialog for Picasso-formatted datasets

A screenshot of the Select Source Files dialog for input of data from Picasso.

The Data Import GUI

The data import GUI displaying a set of localisations (green points) and cell boundaries (blue lines) on top of a reference brightfield image. The user is able to account for spatial offset caused by drift that may occur between the reference image and the fluorescence recording using the blue arrows to the right hand side of the GUI.

The Data Preparaion dialog

The Data Preparation dialog: this interface defines the pixel scale and how tracks should be filtered for length and temporal overlaps. Details of each parameter can be found by hovering over the inputs.

TrackMate-formatted data

TrackMate is a powerful tracking plugin for imageJ/FIJI. While not originally intended for single-molecule tracking, its LoG filter can perform reasonably well for single-molecule tracking data.

Preparation of your data

  • From FIJI/ImageJ, launch the TrackMate plugin, and perform single molecule localisation and tracking using the LoG localisation method (details of how to perform this are available from the TrackMate documentation). Export data using the All spots output option (produces .csv file).
  • Segment cells in your image using the software MicrobeTracker (produces .mat file).

Procedure

  1. Click File > Load new tracking data > TrackMate to launch the Select Source Files dialog (screenshot above, shown for Picasso).
  2. Select the reference image, MicrobeTracker-formatted cell boundaries, TrackMate tracking files, and the fluorescence recording for a single field of view. For multiple recordings see the note below.
  3. Enter an inter-frame time in seconds and press enter; this defines the time between the start of each frame, equal to the exposure time plus any readout time.
  4. Click OK to continue; you will now be presented with the Data Import GUI (screenshot above).
  5. Use the arrows to remove any offset between the localisations and cell boundaries; for some pipelines the definition of y-axis can be inverted, if so click Flip image and cell outlines.
  6. Once aligned click Set static offset and exit.
  7. You will now be presented with Data Preparation GUI (screenshot above); detailed descriptions for each variable can be found by hovering over each input. Click OK.
  8. When prompted, provide a dataset name that will allow to you identify the dataset.
  9. Click OK to continue to the Feature Engineering menu, which is described in the next page of the wiki.

LoColi

LoColi is a non-public single-molecule tracking tool in use by several labs connected to the University of Oxford. Further details will be included here at a later date.

You are now ready to perform feature engineering.

← User guide | Feature engineering →

Clone this wiki locally