Overview#

Basic concepts#

Working with DataLab is very easy. The user interface is intuitive and self-explanatory. The main window is divided into two main areas:

  • The left area contains the Signal Panel and Image Panel tabs, which list the data sets currently loaded in each panel. Clicking a tab switches the main window to the corresponding panel, as well as the menu and toolbar contents. Below the list of data sets, a Properties view shows information about the currently selected data set.

  • The right area shows the visualization of the currently selected data set. The visualization is updated automatically when the user selects a new data set in the list of data sets.

../../_images/s_app_at_startup.en.png

DataLab main window, at startup.#

In the list of data sets, the signal, image or group under the mouse cursor is highlighted, and a delete delete button appears at its right. Clicking this button deletes the item, or the whole selection if the item is part of it, after confirmation. Deleting an item which is not selected keeps the current selection unchanged. Like the “Remove” action of the “Edit” menu, the deletion is recorded in the History Panel. The button may be hidden in the Settings (“Visualization” tab, “Common” sub-tab).

Additional panels are available as dockable widgets, and may be shown or hidden from the “View” menu: the Welcome page, the Applications panel, the History Panel, the Macro panel (see Macros) and the AI Assistant.

Welcome page#

A Welcome page is shown next to the signal and image views when DataLab starts, and whenever the current signal or image panel is empty (for example after switching to an empty panel or deleting its last object). It gathers the main actions to get started:

  • create a signal or an image from a template, open signal or image files, browse an HDF5 file, open a previously saved HDF5 workspace, or import text data (CSV, TSV, …) with the import wizard. When needed, a menu lets you choose between signals and images;

  • ask the AI Assistant, take the guided tour, or run the demo.

Links at the bottom of the page open the online documentation and the release notes of the running version.

An Applications section at the top of the page gathers the application plugins. While no application is installed, it invites you to discover plugins: “Discover plugins…” opens the Catalog tab of the plugin configuration dialog, which lists the plugins of the DataLab plugin catalog, ready to install (see Plugins). This section is hidden when third-party plugins are disabled.

When application plugins are installed, the Applications section shows their tiles, and its “Discover plugins…” button still opens the plugin catalog: clicking the main tile of an application shows its page in the Applications panel, or starts the main action provided by the plugin. An application may provide other tiles, for example to open an example directly: they are shown next to its main tile when the section has room for them, and otherwise move to the menu of the main tile. This menu, opened from the “…” button of the main tile or by right-clicking it, also lets you pin the application to the top of the section or hide it from the welcome page. Pinned applications come first, then recently used ones, then the others by name. The section is limited to two rows of tiles by default (this number may be changed in the Settings): when more applications are installed, the last tile gives the number of applications left out and shows the Applications panel. The “Browse all applications…” button also shows this panel, where the “…” menu of each application offers the same “Show on welcome page” and “Pin to the welcome page” options (see Plugins).

The Welcome page is a regular dockable panel: as soon as a signal or an image is created or opened, the corresponding view is brought to the front, and the Welcome page remains available as a tab. It may be closed like any other panel, and reopened at any time from the “Welcome page” entry of the “?” (Help) menu or from the “View” menu; it is also reopened automatically when the current panel becomes empty. The “Show welcome page when the current panel is empty” check box, at the bottom of the page, controls this automatic display, including at startup (this option is also available in the Settings).

Applications panel#

The Applications panel presents the methods of the installed application plugins. It is shown by the Applications entry of the menu bar, by the application tiles of the Welcome page, or from the “View” menu. The first time, it floats beside the main window, as high as this window, so that you keep your data and their views in sight while using it. You may then dock it like any other panel, for example on the left of the main window, next to the list of data sets whose selection it analyzes, tabify it with another panel, or leave it floating: DataLab shows it again where you left it, with the same size. When the panel is requested while already shown, without switching to another application, it flashes briefly to show where it is.

When several application plugins are installed, their number is given at the top of the panel, above a list which selects the application: click it to show all the applications, with the version and the description of each one. With only one application, its name is shown instead. The “…” button next to it opens a menu with the documentation of the application and its welcome page options: “Show on welcome page” and “Pin to the welcome page”.

The methods of the application are listed one below the other, followed by its tools and datasets, if any: clicking the title of one of these sections hides its content, or shows it again, and DataLab remembers this choice for all applications. Each method takes one row. Its colored dot and the text below its title tell whether the current selection can be analyzed; the tooltip gives the reasons. When metadata are missing, the status points to “Edit > Metadata > Add metadata…”, and it is updated as soon as objects are modified. “Run…” starts the method on the current selection, and asks you to assign the selected objects to its inputs only when needed; the number of created objects and the diagnostics of the run are then shown below the row.

Below the list, in the “Details” area, the panel gives the details of the selected method: click a row, or use the Up and Down keys, to select another method. The details give the version and the description of the method, whether the current selection can be analyzed and why not, the inputs it expects (with their description, and the description and the key of the metadata they require), and the examples designed for it: “Try with this example” opens an example and runs the method with suitable parameters. The splitter between the list and the details sets their heights; DataLab remembers it, as well as the method selected for each application.

Command palette#

Rather than navigating through nested menus, any command may be reached by typing part of its name or of its menu path in the command palette. It lists every command available for the currently active panel, identified by its localised menu path (e.g. “Processing › Fourier analysis › FFT”).

The palette may be opened in three ways:

  • with the Ctrl+Shift+P keyboard shortcut,

  • by clicking the search box in the top-right corner of the menu bar,

  • from the “Command palette…” entry at the top of the “?” (Help) menu.

Type a few characters to filter the list: the matching is fuzzy, so “fft” or “fan” both reach the “Fourier analysis” entries. Results are navigated with Up / Down and triggered with Enter, exactly as if the command had been selected in the menu bar. With an empty query, all commands are listed alphabetically by path.

History Panel#

The History Panel is an additional dockable panel that records signal and image actions in a shared active session, while its Record mode is enabled (by default, this mode is disabled when DataLab starts). It supports replay, step-by-step replay, duplication, and compatibility diagnostics. Sessions may be saved in standalone .dlhist files or with the workspace in HDF5 format.

Internal data model and workspace#

DataLab has its own internal data model, in which data sets are organized around a tree structure. Each panel in the main window corresponds to a branch of the tree. Each data set shown in the panels corresponds to a leaf of the tree. Inside the data set, the data is organized in an object-oriented way, with a set of attributes and methods. The data model is described in more detail in the API section (see sigima.objects).

For each signal or image object, DataLab stores not only the data itself but also a set of metadata describing the data and how it should be displayed. The metadata is stored in a dictionary accessible through the object’s metadata attribute (and may also be browsed in the Properties view, with the Metadata button).

The DataLab Workspace is defined as the collection of all data sets which are currently loaded in DataLab, in both the Signals and Images panels.

Loading and saving the workspace#

The following actions are available to manage the workspace from the File menu:

  • Open HDF5 files…: load a workspace from one or more HDF5 files.

  • Save to HDF5 file…: save the current workspace to an HDF5 file.

  • Browse HDF5 file…: open the HDF5 Browser to explore the content of an HDF5 file and import data sets into the workspace.

Note

Data sets may also be saved or loaded individually, using data formats such as .txt or .npy for 1D signals (see Open signal for the list of supported formats), or .tiff or .png for 2D images (see Open image for the list of supported formats).

Interactive object creation and processing#

DataLab provides an interactive workflow for creating objects and adjusting processing parameters, allowing you to fine-tune results without creating multiple objects.

Interactive object creation#

When creating a new signal or image object using the creation functions (e.g., Gaussian signal, 2D peak image, etc.), DataLab stores the creation parameters in the object’s metadata. This enables interactive parameter adjustment after creation:

  1. Create a signal or image object from the Create menu

  2. Select the created object in the list

  3. A Creation tab appears in the Properties panel (bottom-left)

  4. Modify any creation parameter (amplitude, frequency, size, etc.)

  5. Click Apply to regenerate the object with new parameters

The object is updated in place, preserving any subsequent processing or analysis results. This is particularly useful for:

  • Exploring different parameter values without cluttering the workspace

  • Fine-tuning object characteristics while observing the results

  • Educational purposes to demonstrate parameter effects

Note

Interactive creation is only available for objects created with parameter classes (not for imported data from files).

Interactive 1-to-1 processing#

Live preview before applying a processing#

Standard parameter dialogs for 1-to-1 processing offer a Preview checkbox, unchecked each time the dialog opens. Enable it to display a temporary result while adjusting the parameters. With several signals, images or groups selected, choose one source in the preview selector; OK still processes the whole original selection with the common parameters.

The preview uses the complete source data, including its regions of interest, without adding objects or history entries. Invalid entries suspend updates. Warnings and errors appear inside the dialog. Typing is debounced and slider drags are throttled; large data and expensive algorithms may take longer to update. The preview shows only the result, with its own axes and units, including when an image processing returns a signal.

Starting the first preview creates a dedicated process. After a preview has finished, DataLab keeps this process ready for later preview dialogs, including when switching between signal and image processing. Closing a dialog while a preview is still running stops that process; the next preview then performs a fresh startup. Reloading plugins also discards the cached preview process.

Cancel discards the draft parameters and stops the preview. With one individually selected source, OK reuses a completed, up-to-date preview once and publishes it through the normal processing path, including history. If the preview is still running or stale, or if several objects or groups are selected, DataLab runs the normal processing instead. Analysis, multi-input operations, operations without parameters and unknown custom parameter dialogs retain their existing workflow.

Bounded numeric fields automatically offer sliders in compatible processing forms and in the Processing tab. The exact text field remains authoritative; moving its slider does not limit the precision of values entered by keyboard. Fields without two usable bounds remain text-only. Sliders alone never enable preview or automatic re-processing.

Re-processing an existing result#

When applying a 1-to-1 processing operation that has configurable parameters (e.g., Gaussian filter, threshold, morphological operations), DataLab stores the processing metadata, enabling parameter adjustment and re-processing:

  1. Apply a processing operation with parameters (e.g., Processing > Noise reduction > Gaussian filter)

  2. The result object contains processing metadata (parameters, source object, function name)

  3. Select the processed object in the list

  4. A Processing tab appears in the Properties panel

  5. Modify processing parameters (e.g., filter sigma value)

  6. Click Apply to re-process with updated parameters

Apply recomputes the result in place: the existing object is updated, without creating a new object or a new history entry, and the new execution is recorded in the provenance of the result (see Provenance of results). The parameters of the corresponding History action are left unchanged: to edit them, replay the action with Step-by-step in the History Panel. The optional Auto-recompute on edit checkbox follows the same rules, waits for slider release and valid input, and does not apply edits from a previously selected object. This workflow is ideal for:

  • Iteratively tuning filter parameters while observing results in real time

  • Adjusting threshold values without creating multiple intermediate objects

  • Experimenting with different morphological structuring element sizes

  • Educational demonstrations of parameter effects on processing results

Note

Processing metadata requirements:

  • Processing functions with parameters can be edited; processing functions without parameters (e.g., absolute value, average) offer a Reapply button

  • Every input object must still exist for re-processing (the Processing tab says why it cannot be applied otherwise)

When the workspace holds a provenance record for the object, the Processing tab restores the recorded call, including 2-to-1 processing (source and operand) and n-to-1 processing (every aggregated object, in order). See Provenance of results.

Example workflow#

Here’s a typical workflow using interactive processing:

  1. Create a test signal: Create > Gaussian

    • Default parameters: A=1.0, μ=0, σ=1 (with x from -10 to 10)

  2. Adjust creation parameters: In the Creation tab, change σ to 2, click Apply

    • Signal is regenerated with new width

  3. Apply Gaussian filter: Processing > Noise reduction > Gaussian filter…

    • Default σ=1.0

  4. Fine-tune filtering: In the Processing tab, try σ=3.0, then σ=5.0

    • Each Apply updates the filtered result

    • Compare different smoothing levels without creating multiple objects

This interactive approach provides immediate visual feedback and keeps your workspace organized by avoiding proliferation of test objects with slightly different parameters.