render

Opening Files

Rendering of the super-resolution image: In Picasso: Render, open a movie file by dragging a localization file (ending with ‘.hdf5’) into the window or by selecting File > Open. The super-resolution image will be rendered automatically. A region of choice can be zoomed into by a rectangular selection using the left mouse button. The ‘View’ menu contains more options for zooming and panning.
(Optional) Adjust rendering options by selecting View > Display Settings. The field ‘Oversampling’ defines the number of super-resolution pixels per camera pixel. The contrast settings Min. Density and Max. Density define at which number of localizations per super-resolution pixel the minimum and maximum color of the colormap should be applied.
(Optional) For multiplexed image acquisition, open HDF5 localization files from other channels subsequently. Alternatively, drag and drop all HDF5 files to be displayed simultaneously.

Drift Correction

Picasso offers three procedures to correct for drift: AIM (Ma, H., et al. Science Advances. 2024., option A), use of specific structures in the image as drift markers (option B) and an RCC algorithm (option C). AIM is precise, robust, quick, requires no user interaction or fiducial markers (although adding them will may improve performance). Although RCC does not require any additional sample preparation, option B depends on the presence of either fiducial markers or inherently clustered structures in the image. On the other hand, option B often supports more precise drift estimation and thus allows for higher image resolution. To achieve the highest possible resolution (ultra-resolution), we recommend AIM or consecutive applications of option C and multiple rounds of option B. The drift markers for option B can be features of the image itself (e.g., protein complexes or DNA origami) or intentionally included markers (e.g., DNA origami or gold nanoparticles). When using DNA origami as drift markers, the correction is typically applied in two rounds: first, with whole DNA origami structures as markers, and, second, using single DNA-PAINT binding sites as markers. In both cases, the precision of drift correction strongly depends on the number of selected drift markers.

Adaptive Intersection Maximization (AIM) drift correction

In Picasso: Render, select Postprocess > Undrift by AIM.
The dialog asks the user to select:

Segmentation - the number of frames per interval to calculate the drift. The lower the value, the better the temporal resolution of the drift correction, but the higher the computational cost.

Intersection distance (nm) - the maximum distance between two localizations in two consecutive temporal segments to be considered the same molecule. This parameter is robust, 3*NeNA for optimal result is recommended.

Max. drift in segment (nm) - the maximum expected drift between two consecutive temporal segments. If the drift is larger, the algorithm will likely diverge. Setting the parameter up to 3 * intersection_distance will result in fast computation.

After the algorithm finishes, the estimated drift will be displayed in a pop-up window, and the display will show the drift-corrected image.

Marker-based drift correction

In Picasso: Render, pick drift markers as described in Picking of regions of interest. Use the Pick similar option to automatically detect a large number of drift markers similar to a few manually selected ones.
If the structures used as drift markers have an intrinsic size larger than the precision of individual localizations (e.g., DNA origami, large protein complexes), it is critical to select a large number of structures. Otherwise, the statistic for calculating the drift in each frame (the mean displacement of localization to the structure’s center of mass) is not valid.
Select Postprocess > Undrift from picked to compute and apply the drift correction.
(Optional) Save the drift-corrected localizations by selecting File > Save localizations.

Redundant cross-correlation drift correction

In Picasso: Render, select Postprocess > Undrift by RCC.
A dialog will appear asking for the segmentation parameter. Although the default value, 1,000 frames, is a sensible choice for most movies, it might be necessary to adjust the segmentation parameter of the algorithm, depending on the total number of frames in the movie and the number of localizations per frame. A smaller segment size results in better temporal drift resolution but requires a movie with more localizations per frame.
After the algorithm finishes, the estimated drift will be displayed in a pop-up window, and the display will show the drift-corrected image.

Picking of regions of interest

Manual selection. Open Picasso: Render and load the localization HDF5 file to be processed.
Switch the active tool by selecting Tools > Pick. The mouse cursor will now change to a circle. Alternatively, open Tools > Tools Settings to change the shape into a rectangle. Lastly, choosing Polygon allows for drawing polygons of any shape.
Set the size of the pick circle by adjusting the Diameter field in the tool settings dialog (Tools > Tools Settings). Alternatively, choose Width for a rectangular shape.
Pick regions of interest using the circular mouse cursor by clicking the left mouse button. All localizations within the circle will be selected for further processing.
(Optional) Automated region of interest selection. Select Tools > Pick similar to automatically detect and pick structures that have similar numbers of localizations and RMS deviation (RMSD) from their center of mass than already-picked structures. The upper and lower thresholds for these similarity measures are the respective standard deviations of already-picked regions, scaled by a tunable factor. This factor can be adjusted using the field Tools > Tools Settings > Pick similar ± range. To display the mean and standard deviation of localization number and RMSD for currently picked regions, select View > Show info and click Calculate info below.
(Optional) Exporting of pick information. All localizations in picked regions can be saved by selecting File > Save picked localizations. The resulting HDF5 file will contain a new integer column group indicating to which pick each localization is assigned.
(Optional) Statistics about each pick region can be saved by selecting File > Save pick properties. The resulting HDF5 file is not a localization file. Instead, it holds a data set called groups in which the rows show statistical values for each pick region.
(Optional) The picked positions and diameter itself can be saved by selecting File > Save pick regions. Such saved pick information can also be loaded into Picasso: Render by selecting File > Load pick regions.

NOTE: Rectangular picks can be used to generate the projections of localizations onto the rectangle’s axes. To do so, select rectangular picks and save picked localizations (File > Save picked localizations). The resulting hdf5 files will contain columns x_pick_rot and y_pick_rot, which are the projections of localizations onto and against the “drawing” axis of the rectangle, respectively.

3D rotation window

The 3D rotation window allows the user to render 3D localization data. To use it, select a single pick region (Tools > Pick) and click View > Update rotation window. Some of the display settings (colors, blur method, etc.) are automatically uploaded to the rotation window.

The user may perform multiple actions in the rotation window, including: saving rotated localizations, building animations (.mp4 format), rotating by a specified angle, etc.

Note that to build animations, the user must have ffmpeg installed on their system.

Rotation around z-axis is available by pressing Ctrl/Command. Rotation axis can be frozen by pressing x/y/z to freeze around the corresponding axes (to freeze around the z-axis, Ctrl/Command must be pressed as well).

RESI

In Picasso 0.6.0, a new RESI (Resolution Enhancement by Sequential Imaging) dialog was introduced. It allows for a substantial resolution boost by sequential imaging of a single target with multiple labels with Exchange-PAINT (Reinhardt, Masullo, Baudrexel, Steen, et al., Nature, 2023. DOI: 10.1038/s41586-023-05925-9).

To use RESI, prepare your individual RESI channels (localization, undrifting, filtering and alignment). Load such localization lists into Picasso Render and open Postprocess > RESI. The dialog shown above will appear. Each channel will be clustered using the SMLM clusterer (other clustering algorithms could be applied as well although only the SMLM clusterer is implemented for RESI in Picasso). Clustering parameters can be defined for each RESI channel individually, although it is possible to apply the same parameters to all channels by clicking Apply the same clustering parameters to all channels, which will copy the clustering parameters from the first row and paste it to all other channels.

Next, the user needs to specify whether or not to save clustered localizations or cluster centers from each of the RESI channels individually, and whether to apply basic frame analysis (to minimize the effect of sticking events). For the explanation of the parameters, see SMLM clusterer below.

Upon clicking Perform RESI analysis, each of the loaded channels is clustered, cluster centers are extracted and combined from all RESI channels to create the final RESI file.

G5M

In Picasso 0.9.5, a new algorithm for molecular mapping (i.e., finding the positions of individual molecules from localizations) was introduced: G5M (Gaussian Mixture Modeling with Modifications for Molecular Mapping; Kowalewski, Reinhardt et al. Nature Comms, 2026. DOI: 10.1038/s41467-026-70198-5). G5M is based on Gaussian Mixture Modeling (GMM) but includes several modifications to make it suitable for molecular mapping. All the technicalities as well as the user guide of the method are explained in the publication mentioned and its Supplementary Information. Please refer to picasso.g5m for the details of the implementation. Below is a brief summary of the user guide.

G5M requires some preprocessing of localizations to filter out the badly fitted ones, especially the ones arising from crosstalk (overlapping blinking). These can be excluded from 2D data where the ellipticity and size of the image of an emitter in x and y can be filtered (in Picasso these are found under names “ellipticity”, “sx” and “sy”, respectively). Moreover, the photon count can be cut-off as crosstalk is likely to result in a higher-intensity signal. In 3D data these filters are less reliable due to astigmatism, however, “d_zcalib” could be used. We strongly encourage avoiding dense blinking, where emission signals from neighboring molecules overlap, especially during 3D image acquisition.

Prior to molecular mapping, clustering of localizations is required to split the data into smaller chunks. For many datasets, DBSCAN works well. While in some cases some adjustments may be needed, we recommend the following DBSCAN parameters: In 2D, DBSCAN radius (epsilon) of 2*LP, in 3D - 3*LP (LP - average localization precision of the dataset, for example, NeNA or median localization precision). Default min. samples is set to 4. Clustering in Picasso adds the group column to the localization file, which is required for G5M. Note: G5M relies on the information in the ``group`` column, therefore, if it is overwritten (for example, by picking localizations after DBSCAN clustering), G5M will not work.

To account for fluorophore non-specific sticking, frame analysis is normally recommended (especially the filtering of st. dev. of frame per molecule). However, if localizations from neighboring localization clouds overlap, this is not sufficient due to ambigous assignment of localizations to molecules. Therefore, we recommend filtering of molecules that express too few binding events (saved in the column n_events). In the publication, we recommend a threshold of at least 3 binding events per molecule.

The final postprocessing step is log-likelihood filtering (using the column p_val). The recommended threshold is > 0.0015, however, it might need to be adjusted for your data, especially in 3D this can be too conservative.

As a final check for overfitting (i.e., too many assigned molecules), G5M automatically saves a bar plot of the number of binding events per molecule (n_events column) for clustered (with neighbors within 25 nm) and sparse (without neighbors within 80 nm) molecules. If the clustered molecules show fewer binding events that the sparse molecules, overfitting likely occurred. See Fig. S15 of the publication for an example of well-behaved data. As of v0.9.8, Picasso saves the plot showing relative σ values (i.e., the fitted Gaussian σ divided by the average loc. precision around the molecule). This can be used to estimate if the loc. precision values are accurate (if not, many molecules will have relative σ values close to the user-selected min./max. σ). As of version 0.9.10, Picasso runs KS 2 sample test to compare the two distributions. The output test statistic and theoretical p value correspond to the KS test, while the permutation p value is calculated by randomly permuting the labels of clustered and sparse molecules 1,000 times and calculating the fraction of permutations that result in a KS test statistic as extreme as the one observed with the original labels.

If the outcome of G5M seems unsatisfactory, please check the following:

Make sure that group column is present in the localization file and contains the correct information (i.e., from DBSCAN clustering, not from picking localizations);
Make sure that the loc. precision values (columns lpx, lpy, lpz) are correct, comparing NeNA and median loc. precision is a reasonable proxy (without fiducial markers); the most common issue is a miscalibrated camera, leading to incorrect photon counts and thus incorrect loc. precisions;
Another reason why the loc. precision values can be off is due to the small box size in the localization step; especially in 3D astigmatic imaging, single-emitter images can be quite large, potentially exceeding the user-defined box size; in such cases, we recommend increasing the box size in the localization step and rerunning the analysis;
Inspect if the localizations were preprocessed as described above;
Rerun the analysis without postprocessing (filtering) and redo it manually, since some steps may be too stringent, such as p_val or n_events (latter especially for short acquisition times);
Adjust min./max. σ, especially too low max. σ may lead to high false positive error rates (i.e., overfitting); We suggest inspecting rel_sigma values of the assigned molecules, which are calculated as the fitted σ divided by the mean localization precision of the surrounding localizations. If the values are close to the user-selected min./max. σ, min./max. σ might need to be adjusted. Alternatively, this might be a sign of inaccurate/inprecise loc. precision values, see above;
Adjust min. locs;
Adjust DBSCAN (or other clustering algorithm) parameters. For example, if G5M takes too long to run, the DBSCAN clusters most likely contain too many molecules. In such a case, we recommend splitting such clusters further;

Dialogs

Display Settings

Allows to change the display settings. Open via View > Display Settings.

General

Adjust the general display settings.

Zoom

Set the magnification factor.

Oversampling

Set the oversampling. Choose dynamic to automatically adjust to current window size when zooming.

Minimap

Click show minimap to display a minimap in the upper left corner to localize where the current field of view is within the image.

Contrast

Define the minimum and maximum density of the and select a colormap. Over 100 colormaps are available. The last option Custom requires the user to load their own .npy file containg a numpy array with a custom colormap. The selected colormap will be saved when closing render.

Blur

Select a blur method. Available options are: * None * One-Pixel-Blur * Individual Localization Precision * Individual Localization Precision, iso

Camera

Select the pixel size of the camera. This will be automatically set to a default value or the value specified in the *.yaml file.

Scale Bar

Activate scale bar. The length of the scale bar is calculated with the Pixel Size set in the Camera dialog. Activate Print scale bar length to additionally print the length.

Render properties

This allows rendering properties by color.

Show Info

Displays the info dialog.

Display

Shows the image width/height, the coordinates, and dimensions of the current FoV.

Movie

Displays the median fit precision of the dataset. Clicking on Calculate allows calculating the precision via the NeNA approach. See DOI: 10.1007/s00418-014-1192-3.

FRC

Displays the FRC resolution of the dataset. Takes in the image in the current FOV and calculates the FRC resolution via splitting the localizations into two halves. Based on the approach from 10.1038/nmeth.2448. Does not take into account the Q factor for multiple blinking.

Field of view

Shows the number of localizations in the current FoV.

Picks

Allows calculating statistics about the picked localizations. Press Calculate info below to calculate. Ignore dark times allows treating consecutive localizations as on, even if there are localizations (specified by the parameter) missing between them. When defining the number of units per pick, you can calibrate the influx rate via Calibrate influx. A histogram of the dark and bright time can be plotted when clicking Histograms.

Menu items

File

Open [Ctrl+O]

Open an .hdf5 file to open in render.

Open rotated localizations [Ctrl+Shift+O]

Opens localizations that were saved via the rotation window, see above.

Save localizations [Ctrl+S]

Save the localizations that are currently loaded in render to an hdf5 file.

Save picked localizations [Ctrl+Shift+S]

Save the localizations that are within a picked region (yellow circle, rectangle or polygon). Each pick will get a different group number. To display the group number in Render, select Annotate picks in Tools/Tools Settings. In case of rectangular picks, the saved localizations file will contain new columns x_pick_rot and y_pick_rot, which are localization coordinates into the coordinate system of the pick rectangle (coordinate (0,0) is where the rectangle was started to be drawn, and y_pick_rot is in the direction of the drawn line.) These columns can be used to plot density profiles of localizations along the rectangle dimensions easily (e.g., with “Filter”).

Save pick properties

Calculates the properties of each pick (i.e., mean frame, mean x mean y as well as kinetic information) and saves it as an hdf5 file.

Save pick regions

Saves the positions of the picked regions (yellow circles) in a .yaml file. The file will contain the following: A list of center positions and the value of the diameter. It is possible to manually add center positions or copy from another pick regions file with a text editor.

Load pick regions

Resets the current picked regions and loads regions from a .yaml file that contains pick regions.

Export ROI for Imaris

This function allows to export the current ROI for Imaris. Note that this is currently only implemented for Windows. Click on File / Export ROI for imaris and enter a filename for export. Picasso will export the current region of interest with the current oversampling settings. If multiple channels are loaded it will export the channels with the same colors as set in Picasso (Shortcut CTRL+F or View / Files to change.) Depending on the size of the ROI, the export will take a couple of seconds. Once exporting is finished, the file will be saved at the set location. The resulting file can be opened e.g. with ImarisViewer or Imaris. Note that the orientation is the same as in Picasso.

Export localizations

Select export for various other programs. Note that some exporters only work for 3D files (with z coordinates). For additional file converters check out the convert folder at Picasso’s GitHub page.

Export as .csv for ThunderSTORM

This will export the dataset in a .csv file to use with ThunderSTORM.

Note that for large datasets the writing of the file may take some time.

Note that the pixel size value that is set in Display Settings will be used for exporting.

Thefollowing columns will be exported: 3D: id, frame, x [nm], y [nm], z [nm], sigma1 [nm], sigma2 [nm], intensity[photon], offset[photon], uncertainty_xy [nm] 2D: id, frame, x [nm], y [nm], sigma [nm], intensity [photon], offset [photon], uncertainty_xy [nm]

The uncertainty_xy is calculated as the mean of lpx and lpy. For 2D, sigma is calculated as the mean of sx and sy.

For the case of linked localizations, a column named detections will be added, which contains the len parameter - that’s the duration of a blinking event and not the number n of linked localizations. This is meant to be better for downstream kinetic analysis. For a gradient that is well-chosen n ~ len and for a gap size of 0 len = n.

Export as .txt for FRC

Export as .txt file to be used for the fourier ring correlation plugin in ImageJ.

Export as .xyz for Chimera

Export as .txt file to be used for Chimera import.

Export as .3d for ViSP

Export as .3d file to be used ViSP.

Remove all localizations

Removes all .hdf5 files loaded, restarts the render window.

View

Display settings (CTRL + D)

Opens the Display Settings Dialog.

Files (CTRL + F)

Opens the Datasets dialog, which lists every loaded channel and lets you control its title, visibility, color (or colormap), and relative intensity. A small horizontal gradient next to each channel previews what that channel will look like at intensity 0 → intensity 1.

Each channel’s Color dropdown is organised into three sections:

Solid colors — the 14 default named colors (red, cyan, green, …). You can also type a hexadecimal code such as #FF5733 directly into the dropdown. Solid colors are rendered as a black → color ramp, exactly matching the previous “intensity × RGB” behaviour.
Built-in colormaps — one 3-stop black → color → white gradient per default solid color, named <color>_gradient (e.g. blue_gradient, red_gradient).
Custom — any user-defined colormaps (see “Edit custom colormaps…” below). This section only appears once at least one custom colormap has been defined.

Channels are blended additively in the final image and clipped to 1.0, so overlapping high-intensity regions saturate toward the sum of the channel colors.

The Automatic coloring checkbox overrides per-channel selections with HSV-spaced colors for as long as it’s ticked. Save colors / Load colors write / read a one-identifier-per-line .txt file — any name from the three dropdown sections (or a hex code) is valid.

Edit custom colormaps

Opens a small editor where you can create, rename, duplicate, or delete your own colormaps. Each custom colormap is a list of 2-5 stops; each stop has a position in [0, 1] and an RGB color. Stops are linearly interpolated into the 256-row look-up table (LUT) used at render time. Click any of the R / G / B cells to type a value, or double-click the row to pick the stop color from a standard color dialog. Use Add stop / Remove stop to grow or shrink the gradient.

Programmatic use

The underlying conversion from solid colors or stops to a (256, 3) LUT is also exposed as part of picasso.render:

from picasso import render
lut_red   = render.solid_to_lut((1.0, 0.0, 0.0))     # black → red
lut_fire  = render.stops_to_lut([(0, 0, 0, 0),
                                 (0.5, 1, 0, 0),
                                 (1, 1, 1, 0)])      # black → red → yellow
qimage, *_ = render.render_scene(
    locs=..., info=..., colors=[lut_red, lut_fire], ...
)

Passing a list of LUTs to render_scene selects the per-channel colormap path; passing a list of plain RGB triplets (legacy) still works and is equivalent to solid_to_lut per channel.

Left / Right / Up / Down

Moves the current field of view in a particular direction. Also possible by using the arrow keys.

Zoom in (CTRL +)

Zoom into the image.

Zoom out (CTRL -)

Zoom out of the image.

Fit image to window

Fits the reconstructed image to be fully displayed in the window.

Slice (3D)

Opens the slicer dialog which allows for slicing through 3D datasets.

Update rotation window (3D) [Ctrl+Shift+R]

Opens/updates rotation window, see above. Requires a single picked region of interest to be selected.

Show info

Shows info for the current dataset. See Info Dialog.

Tools

Zoom (CTRL + Z)

Selects the zoom tool. The mouse can now be used for zoom and pan.

Pick (CTRL + P)

Selects the pick tool. The mouse can now be used for picking localizations. The user can set the pick shape in the Tools settings (CTRL + T) dialog. The default shape is Circle with the diameter to be set. For rectangles, the user draws the length, while the width is controlled via a parameter for all drawn rectangles, similar to the diameter for circular picks. For a polygonal pick, the user clicks with the left button to draw the desired polygon. The right button deletes the last selected vertex. The polygon can be close by clicking with the left button on the starting vertex.

Measure (CTRL + M)

Selects the measure tool. The mouse can now be used for measuring distances. Left click adds a crosshair for measuring; right-click deletes the last crosshair.

Tools settings (CTRL + T)

Define the settings of the tools, i.e., the radius of the pick and an option to annotate each pick. For the circular picks the range of pick similar can be set.

Pick similar (CTRL + Shift + P)

Automatically identifies picks that are similar to the current picks.

Remove localizations in picks

Remove localizations found in picked region(s) of interest. Can be applied to separate or all channels simultaneously.

Move to pick

Changes FoV to display a pick region specified by the user.

Pick fiducials

Automatically picks fiducials. To do so, the whole FOV image is rendered at one-pixel-blur. Then, such image pixel intesities are histogramed and the 99th is used as a threshold for selecting image maxima using Localize’s identification.

Show trace (CTRL + R)

Shows the time trace of the currently selected pick(s).

Select picks (trace)

Opens a dialog to that goes through all picks, displays its trace and asks to keep or discard it.

Select picks (XY scatter)

Opens a dialog to that goes through all picks, displays a xy-scatterplot and asks to keep or discard it.

Plot pick (XYZ scatter) (CTRL + 3)

Displays a 3D scatterplot of the localizations of the currently selected pick(s).

Select picks (XYZ scatter)

Opens a dialog to that goes through all picks, displays an xyz-scatterplot and asks to keep or discard it.

Select picks (XYZ scatter, 4 panels)

Opens a dialog to that goes through all picks, displays four panels with an xyz-scatterplot and a top, bottom and side projection and asks to keep or discard it.

Filter picks by locs

Allows filtering picks by the number of localizations in each pick. When clicking, a histogram of the number of localizations of all selected picks will be calculated. A lower and upper boundary can be selected to filter the picks.

Clear picks (Ctrl + C)

Clears all currently selected picks.

Subtract pick regions

Allows loading another pick regions file to subtract from the currently selected picks. Can be slow for a large number of picks.

Cluster in pick (k-means)

Allows performing k-means clustering in picks. Users can specify the number of clusters and deselect individual clusters. Picks can be kept or removed. After looping through all picks an hdf5 file with the cluster information can be saved.

Mask image

Opens a dialog that allows the user to specify a mask for filtering localizations within and outside it. The user can adjust the histogram bin size, blur thereof and the threshold applied.

The images can be zoomed in/out (Ctrl/Cmd + scrolling) and panned (mouse right click). Double clicking resets the zoom.

Fast rendering

Allows the user to display only a fraction of localizations to speed up rendering.

Postprocess

Undrift by AIM

Performs drift correction using the AIM algorithm (Ma, H., et al. Science Advances. 2024).

Undrift from picked (3D)

Performs drift correction using the picked localizations as fiducials. Also performs drift correction in z if the dataset has 3D information.

Undrift from picked (2D)

Performs drift correction using the picked localizations as fiducials. Does not perform drift correction in z even if dataset has 3D information.

Undrift by RCC

Performs drift correction by redundant cross-correlation.

Undo drift (2D)

Undo previous drift correction (only 2D part). Can be pressed again to redo.

Show drift

After drift correction, a drift file is created. If the drift file is present, the drift can be displayed with this option.

Apply drift from an external file

Applies drift from a user-specified. txt file. Keep in mind that the .txt drift files after consecutive undrifting rounds produce cumulative drift. Therefore, if 3 rounds of undrifing were performed, only the last file specifies the drift calculated in the 3 steps.

Remove group info

Removes the group information when loading a dataset that contains group information. This will, i.e., turn the multicolor representation into a single color representation.

Sync groups across channels

If more than one channel is present, this function rejects localizations from across the channels whose group field is not present in all channels. This is useful for removing, for example, clustered localizations after their cluster centers were filtered with frame analysis.

Unfold / Refold groups

Allows to “unfold” an average to display each structure individually in a line.Note that the structures need to be grouped and processed with Picasso: Average beforehand.

Unfold groups (square)

Arranges an average in a square so that each structure is displayed individually. This function does not require Picasso: Average beforehand. Instead, grouped or picked (circular picks) localizations are accepted.

Link localizations

Links consecutive localizations

Align channels (RCC or from picked)

Aligns channels to each other when several datasets are loaded. If picks are selected, the alignment will be via the center of mass of the picks; otherwise, an RCC will be used.

Combine locs in picks

Combines all localizations in each pick to one.

Apply expressions to localizations

This tool allows you to apply expressions to localizations, for example:

x +=1 will shift all localization by one to the right
x +=1; y+=1 will shift all localization by one to the right and one up.
flip x z will exchange the x-axis with y-axis if z localizations are present (side projection), similar for flip y z.
spiral r n will plot each localization over the time of the movie in a spiral with radius r and n number of turns (e.g., to detect repetitive binding), uspiral to reverse.

NOTE: using two variables in one statement is not supported (e.g. x = y) To filter localizations use picasso filter.

DBSCAN

Cluster localizations with the dbscan clustering algorithm.

HDBSCAN

Cluster localizations with the hdbscan clustering algorithm.

SMLM clusterer

Cluster localizations with the custom algorithm designed for SMLM. In short, localizations with the maximum number of neighboring localizations within a user-defined radius are chosen as cluster centers, around which all localizations within the given radius belong to one cluster. If two or more local maxima are within the radius, the clusters are merged.

SMLM clusterer requires three (or four if 3D data is processed) arguments:

Radius: final size of the clusters.
Radius z (3D only): final size of the clusters in the z axis. If the value is different from radius in xy plane, clusters have ellipsoidal shape. Radius z can have a different value to account for a difference in localization precision in lateral and axial directions.
Min. locs: minimum number of localizations in a cluster.
Basic frame analysis: If True, each cluster is checked for its value of mean frame (if it is within the first or the last 20% of the total acquisition time, it is discarded). Moreover, localizations inside each cluster are split into 20 time bins (across the whole acquisition time). If a single time bin contains more than 80% of localizations per cluster, the cluster is discarded.

Note to all clustering algorithms: it is highly recommended to remove any fiducial markers before clustering, to lower clustering time, given they are of no interest to the user. To do that, the markers can be picked and removed using Tools > Remove localizations in picks.

Test clusterer

Opens a dialog where different clustering parameters can be checked on the loaded dataset. Requires a single pick region of interest to be selected.

Nearest Neighbor Analysis

Calculates distances to the k-th nearest neighbors between two channels (can be the same channel). k is defined by the user. The distances are stored in nm as a .hdf5 localizations file with new columns nnd_1, nnd_2, …, nnd_k for each localization in channel 1. The distances are calculated in 3D if both datasets have z information.