Dataset DOI: 10.5061/dryad.d2547d8jx

Description of the data

The dataset includes files of 24 hour gastruloids from different initial cell proportions. The data corresponds to different experimental replicates and corresponding figures in the manuscript D. Oriola, G. Torregrosa-Cortes et al., bioRxiv, 2026:

- Analysis_data_24h_unmixing_210610.zip : Experiment 210610. (Fig. S7)

- Analysis_data_24h_unmixing_211116.zip : Experiment 211116. (Fig. 3C,D and Fig. S7)

- Analysis_data_24h_unmixing_220510.zip: Experiment 220510. (Fig. S7)

The data was acquired using a PerkinElmer Opera Phenix system using a x20 dry objective and N.A. 0.4 in confocal mode. The dataset consists of (i) raw image files .tif and (ii) files generated after the segmentation analysis (.csv, .pkl, .json) using the software MOrgAna. The model folder contains two subfolders (trainingset and result_segmentation) and the files describing the parameters of the segmentation model. In the first subfolder the ground truth images can be found. The second folder contains the final masks, the quantification on gastruloid morphology and intensity of the fluorescence signals. 

Here we detail the different types of files:

(i) Raw image files

The snapshots correspond to images of gastruloids in 96 well plates, where each well contains a single gastruloid. Each well is identified with an alphanumerical code (e.g. 01F meaning row 1, column F). 

• Brightfield and GFP channel images (BF_GFP_well.tif): Snapshots with the brightfield images and the GFP fluorescence signal coming from Bra/T+ cells in the aggregate. The 
• Brightfield and SiR-DNA images (BF_sirDNA_well.tif): Snapshots with the brightfield images and the fluorescence signal coming from the SiR-DNA dye used to label a fraction of cells. 
• Ground truth images (BF_sirDNA_well_GT.tif): Images selected as the trainingset for segmentation. They are found in the trainingsetfolder.

(i) Segmentation files (output from the software MOrgAna)

• Final masks: Files ending with _finalMask.tif  inside the result_segmentation folders correspond to the final masks obtained after segmentation. Those can be either watershed, classifier or manual masks.
• Summary of final masks: Files ending with _finalMasks.pnginside the result_segmentation folders correspond to a summary of all the final masks obtained after segmentation.
• Watershed masks: Files ending with _watershed.tif  inside the result_segmentation folders correspond to the watershed masks.
• Classifier masks: Files ending with _classifier.tif  inside the result_segmentation folders correspond to the classifier masks.
• Manual masks: Files ending with _manual.tif  inside the result_segmentation folders correspond to the manual masks.
• Fluorescence intensity analysis: Files ending with _fluo_intensity.json inside result_segmentation folders correspond to the output on fluorescence quantification (see MOrgAna for details)
• Morphological analysis: Files ending with _morpho_params.json inside result_segmentation folders correspond to the output on morphology quantification (see MOrgAna for details)
• Segmentation parameters: The file segmentation_params.csv is generated by MOrgAna and contains all the segmentation information.  
• Classifier: The trained classifier corresponds to the file classifier.pkl
• Feature scaler: The scaler file scaler.pkl stores the training-set values so that new images are transformed in exactly the same way before classification.
• Parameters of the segmentation model:  The file params.json stores information needed to reconstruct the model or preprocessing pipeline

Usage notes to access the different files

• .tif files: They can be opened using Fiji/ImageJ software.  

• .csv files: They can be opened using Excel or a Text Editor. In Python you can use: 

  import pandas as pd

  df = pd.read_csv("data.csv")

  print(df.head())

• .json files: They can be accessed using a Text Editor. In Python you may use:

  import json

  with open("data.json", "r") as f:

          data = json.load(f)

  print(data)

• .pkl files: They can be accessed with Python using:

  import pickle

  with open("data.pkl", "rb") as f:

               data = pickle.load(f)

  print(data)

• .png files: They can be accessed using Preview on Mac, Photos, a web browser, etc

Instructions for data analysis procedure

The dataset can be analyzed using the Python files (.py) found in the Radial_analysis folder from the Github repository https://github.com/MPoLS-lab/gastrusym.