DNAsight

DNAsight is a modular framework for automated segmentation and quantitative analysis of AFM images of DNA–protein complexes. See associated publication / for citation: https://doi.org/10.64898/2026.03.06.708946

It provides both:

• DNAsight GUI for interactive, point-and-click analysis
• DNAsight CLI for batch processing and reproducible pipelines

---

Contents

• Installation
  • Option 1 — Install with pip (recommended if you have Python)
  • Option 2 — Download the standalone app (no Python required)
• Quick start
• Using the GUI
• Using the CLI
• Configuration
• Example data
• Outputs
• Troubleshooting
• Development
• Citation
• License
• Contact

---

Installation

DNAsight can be installed in two ways:

1. Python (pip) installation (recommended for users who already have Python, e.g. MacOS users)
2. Standalone app download (recommended if you don’t want to install Python)

Option 1 — Install with pip (recommended if you have Python)

This option installs DNAsight as a Python package and is the easiest way to stay up to date. It also avoids macOS “unblock” steps that can occur when running downloaded applications.

Prerequisites

• Python (recommended: Python 3.10+)
• A clean virtual environment (strongly recommended)

1) Create and activate a clean virtual environment

macOS / Linux

python3 -m venv dnasight-env
source dnasight-env/bin/activate

Windows (PowerShell)

python -m venv dnasight-env
.\dnasight-env\Scripts\Activate.ps1

2) Install DNAsight

pip install dnasight

3) Launch DNAsight

GUI:

dnasight-gui

CLI:

dnasight-cmd --help

If the commands aren’t found, make sure your environment is activated.

---

Option 2 — Download the standalone app (no Python required)

This option is best if you don’t want to install Python. Download the correct release for your operating system, unzip it, then run the GUI or CLI.

1) Download the release

• Go to the GitHub Releases page
• Download the macOS or Windows zip (choose the one that matches your system)

2) Unzip

Unzip the downloaded file. You should see a folder containing something like:

• dnasight-gui (GUI application)
• dnasight-cmd (command-line executable)
• data/ (example datasets to test the workflow)
• config.yaml (configuration template)

3) Run DNAsight

GUI

• Double click dnasight-gui to open DNAsight.

CLI

Open a terminal in that folder and run the command-line executable.

macOS (example):

./dnasight-cmd --help

Windows (example, PowerShell):

.\dnasight-cmd.exe --help

Replace the commands above with the exact executable name on each platform if it differs.

macOS note: “Unblock” on first run

If macOS blocks the app the first time you try to open it:

1. Go to System Settings → Privacy & Security
2. Find the DNAsight warning
3. Click Open Anyway

---

Quick start

This works for both installation methods.

1. Launch the GUI (dnasight-gui)
2. Load an image or folder (or use the example files in data/)
3. Run segmentation and analysis
4. Export results (CSV + overlays)

---

Using the GUI

Typical workflow:

1. Choose segmentation and quantification modules, carefully adjusting settings to your experiment
2. Select an input folder containing images in TIFF format and select fixed pixel size or generate and fill out pixel size csv
3. Set calibration options
4. Run DNAsight (press "Run" to start and "Stop" to stop the analysis before it finishes).
5. Check results in output folder

---

Using the CLI

The CLI enables batch processing and reproducible runs.

Show CLI help:

dnasight-cmd --help

---

Configuration

DNAsight can be configured with config.yaml (included in the standalone zip). This file stores defaults for common settings.

• If you are using the standalone app, you will see config.yaml next to the executables.
• If you installed via pip, DNAsight uses internal defaults and (optionally) a user-provided config.

Add details here about where config is read from (working directory vs user directory) once finalized.

---

Example data

The standalone download includes a data/ folder with example datasets you can use to test the full workflow.

---

Outputs

DNAsight generates csvs and/or image overlays and plots depending on the applied modules.

S1: DNA segmentation

The DNA segmentation module outputs one .tif file for each input image containing two layers:

1. the raw image
2. the segmentation

The segmentation layer also contains the global ID of each DNA molecule in the individual masks. The output .tif files have a_ added in front of the original filename to indicate that they are annotated files.

In addition, a folder called segmentation_plots is generated. This folder contains .png files with four plots for each input image and is intended to help the user assess DNA segmentation quality. These files are named in the same way as the annotated outputs.

S2: Cluster segmentation

The cluster segmentation module outputs inside folder cluster_segmentation:

• an .npy file containing the segmentation mask and global ID for each cluster
• a .png file intended to help users assess whether the cluster segmentation parameters should be adjusted
• a .csv file called segmentation_results
• a separate .csv file called cluster_quantification

segmentation_results.csv

This file contains the following columns:

• global_cluster_id: global ID of each cluster within the provided files. If you run a different folder, the IDs may repeat. They are only global within the given folder.
• file: filename of the corresponding .tif file
• local_id: local cluster ID within the image file, used for internal reference for downstream quantifications
• centroid_x: x coordinate of the centroid of a detected cluster in pixels
• centroid_y: y coordinate of the centroid of a detected cluster in pixels
• intensity: summed intensity of pixels in the segmented cluster
• area: area of the segmented cluster in pixels

Q1: DNA length calculation

The DNA length calculation module outputs a folder called overlays, which contains .png files for each input image with the length of each segmented DNA component overlaid. The files are named as the annotated files, that is, with a_ added in front of the original filename.

It also outputs a .csv file called length_per_component, which contains information about the length of each segmented DNA component.

length_per_component.csv

This file contains the following columns:

• filename: filename of the annotated file
• comp_id: global ID of each DNA molecule within the provided files. If you run a different folder, the IDs may repeat. They are only global within the given folder.
• area_px: area of the segmented DNA molecule in pixels
• touches_edge_dna: True or False depending on whether the segmented DNA molecule touches the image edge
• length_px: length of the segmented DNA molecule in pixels
• length_nm: length of the segmented DNA molecule in nm
• length_bp: length of the segmented DNA molecule in bp
• length_bp_sem: estimated error on the length of the segmented DNA molecule in bp based on the provided calibration
• had_valid_pixel_size: True or False indicating whether a valid pixel size was given for conversion
• bp_calibration_used: True or False indicating whether a calibration was applied

Q2: DNA spatial organization

The DNA spatial organization module outputs a .csv file called geometric_features_summary, which contains information about the extracted geometric features. It also outputs .pdf files with selected geometric feature values overlaid the raw images.

geometric_features_summary.csv

This file contains the following columns:

• filename: filename of the annotated file
• pixel_size: pixel size of the given file, if provided
• comp_id: global ID of each DNA molecule within the provided files. If you run a different folder, the IDs may repeat. They are only global within the given folder.
• touches_edge_dna: True or False depending on whether the segmented DNA molecule touches the image edge
• length_px: length of the segmented DNA molecule in pixels
• length_nm: length of the segmented DNA molecule in nm
• rg_px: radius of gyration of the segmented DNA molecule in pixels
• rg_nm: radius of gyration of the segmented DNA molecule in nm
• compaction: reciprocal of the radius of gyration normalized to the length of the DNA molecule; unitless
• n_branch_clusters: number of crossings of the segmented DNA molecule
• tortuosity: tortuosity of the segmented DNA molecule; empty if fewer than one end was found, for example for plasmids
• elongation: elongation metric
• mean_kappa_px_inv: mean curvature of the segmented DNA molecule in pixel space
• std_kappa_px_inv: standard deviation of curvature of the segmented DNA molecule in pixel space
• min_kappa_px_inv: minimum curvature of the segmented DNA molecule in pixel space
• max_kappa_px_inv: maximum curvature of the segmented DNA molecule in pixel space
• mean_kappa_nm_inv: mean curvature of the segmented DNA molecule in real space, in 1/nm
• std_kappa_nm_inv: standard deviation of curvature of the segmented DNA molecule in real space, in 1/nm
• min_kappa_nm_inv: minimum curvature of the segmented DNA molecule in real space, in 1/nm
• max_kappa_nm_inv: maximum curvature of the segmented DNA molecule in real space, in 1/nm
• no_strong_bends_px: number of strong bends evaluated over a given pixel distance
• no_strong_bends_nm: number of strong bends evaluated over a given real-space distance in nm
• lp_px: persistence length in pixel space
• lp_nm: persistence length in real space, in nm
• error: an error message appears here if the calculation of a feature failed

Q3: DNA loop quantification

The DNA loop quantification module outputs a folder called loops, which contains one .pdf file for each input image. Each .pdf shows detected loops overlaid with loop length and loop position given in pixels.

A yellow circle is used to indicate anchor points for loops where the intensity of the overlap exceeds the mean plus standard deviation of the intensity of the DNA molecule.

The module also outputs a csv file named loops_summary.csv which contains information about each loop.

Loop output loops_summary.csv

This file contains the following columns:

• filename: filename of the annotated file
• comp_id: global ID of each DNA molecule within the provided files. If you run a different folder, the IDs may repeat. They are only global within the given folder.
• loop_index: loop number within the image
• loop_length_px: length of the loop in pixels
• raw_dist_px: distance from the closest end to the loop, excluding loops along the way, in pixels
• dist_incl_loops_px: distance from the closest end to the loop, including loops along the way, in pixels
• n_loops_added_inclusive: number of other detected loops in the same DNA object that were added to the path length because they lie on the shortest path from a skeleton endpoint to the loop attachment point
• n_loops_excluded_samepos: number of other loops not counted because they are effectively at the same loop position as the current loop
• mean_dna_intensity: mean intensity of the full DNA molecule
• std_dna_intensity: standard deviation of intensity of the full DNA molecule
• mean_anchor_intensity: mean intensity at the loop anchor position
• std_anchor_intensity: standard deviation of intensity at the loop anchor position
• loop_length_nm: length of the loop in nm; empty if no pixel size is provided
• raw_dist_nm: distance from the closest end to the loop, excluding loops along the way, in nm; empty if no pixel size is provided
• dist_incl_loops_nm: distance from the closest end to the loop, including loops along the way, in nm; empty if no pixel size is provided
• loop_length_bp: length of the loop in bp; empty if no calibration is provided
• raw_dist_bp: distance from the closest end to the loop, excluding loops along the way, in bp; empty if no calibration is provided
• dist_incl_loops_bp: distance from the closest end to the loop, including loops along the way, in bp; empty if no calibration is provided
• touches_edge_dna: True or False depending on whether the segmented DNA molecule touches the image edge
• attachment_x: x coordinate of the loop anchor in pixels
• attachment_y: y coordinate of the loop anchor in pixels

Q4: Protein-DNA associations

The protein-DNA association quantification module outputs the inside folder cluster_quantification, where it creates a folder named overlays, which saves .pdf files showing overlays of associated clusters and DNA (in bp). Inside cluster_quantification it outputs the followings csvs:

cluster_quantification.csv

This file contains the following columns:

• cluster_id: global ID of each cluster within the provided files. If you run a different folder, the IDs may repeat. They are only global within the given folder.
• file: filename of the corresponding .tif file
• summed_intensity_px: summed intensity of pixels within the cluster segmentation
• bg_ring_mean: local background around the segmented cluster
• bg_corrected_summed_intensity_px: background-corrected summed intensity of pixels
• touches_edge_cluster: True or False depending on whether the segmented cluster touches the image edge
• centroid_x: x coordinate of the centroid of a detected cluster in pixels
• centroid_y: y coordinate of the centroid of a detected cluster in pixels
• pixel_size: pixel size of the given image when applied; empty if no pixel size is given
• cluster_area_nm2: area of the segmented cluster in nm²; empty if no pixel size is given
• summed_intensity_per_nm: summed intensity of the segmented cluster normalized to pixel size; empty if no pixel size is given
• summed_intensity_per_nm2: summed intensity of the segmented cluster normalized to area; empty if no pixel size is given
• bg_summed_intensity_per_nm: local background-corrected summed intensity of the segmented cluster normalized to pixel size; empty if no pixel size is given
• bg_summed_intensity_per_nm2: local background-corrected summed intensity of the segmented cluster normalized to area; empty if no pixel size is given

group_summary.csv

This file contains one row per connected group of clusters (i.e. clusters connected either directly or via shared DNA).

• filename: filename key of the corresponding image
• n_clusters_in_group: number of clusters in the group
• cluster_ids: list of global cluster IDs belonging to the group
• dna_ids: list of DNA molecule IDs associated with the group
• total_length_px: total DNA length associated with the group in pixels (sum over all associated DNA molecules)
• total_length_nm: total DNA length in nanometers; empty if no pixel size is given
• total_length_bp: total DNA length in base pairs; empty if no calibration is available
• lengths_px_list: list of individual DNA segment lengths (in pixels) associated with the group
• lengths_nm_list: list of individual DNA segment lengths (in nanometers); empty if no pixel size is given
• lengths_bp_list: list of individual DNA segment lengths (in base pairs); empty if no calibration is available

This file includes all DNA segments associated with each group, including terminal segments at DNA ends. If desired, the output can be filtered downstream (e.g. using the DNA-centered summary) to include only DNA segments associated with a specific number of clusters, such as exactly two clusters to focus on linker-like segments between nucleosomes.

dna_centered_summary.csv

This file contains one row per DNA molecule that is associated with at least one cluster.

• filename: filename key of the corresponding image
• dna_id: global ID of the DNA molecule within the image
• n_clusters_associated: number of clusters associated with the DNA molecule
• cluster_ids: list of global cluster IDs associated with the DNA molecule
• length_px: DNA length in pixels; empty if the DNA length was not found
• length_nm: DNA length in nanometers; empty if no pixel size or calibration is available
• length_bp: DNA length in base pairs; empty if no base-pair calibration is available

This file enables filtering of DNA molecules based on their cluster associations. For example, selecting DNA molecules with n_clusters_associated == 2 restricts the analysis to DNA segments associated with exactly two clusters, which can be used to isolate linker-like DNA segments between nucleosomes.

Notes

• IDs described as global are only global within a single run or folder.
• If another folder is processed separately, IDs may repeat.
• Several outputs depend on whether pixel size and bp calibration were supplied.
• Empty fields generally indicate that the required metadata was not available or that the calculation was not applicable.

---

Troubleshooting

macOS: app is blocked on first open

• Go to System Settings → Privacy & Security
• Click Open Anyway for DNAsight

CLI command not found (pip install)

• Make sure your virtual environment is activated
• Reinstall with pip install dnasight inside the environment

Still stuck?

Open an issue on GitHub with:

• your OS version
• how you installed DNAsight (pip vs standalone)
• the exact error message

---

Development

For development installs (contributors):

git clone https://github.com/emilywinther/dnasight
cd dnasight
pip install -e .

---

License

MIT

---

Contact

For questions or bug reports, please open a GitHub issue.