Dynamic altas of early Drosophila development created using in-toto lightsheet microscopy and tissue cartography.

Reference Information

Dataset contributors

Mitchell, Noah, University of California, Santa Barbara
Lefebvre, Matthew, University of California, Santa Barbara
Jain-Sharma, Vishank, University of California, Santa Barbara
Claussen, Nikolas, University of California, Santa Barbara
Raich, Marion, Technical University Munich
Gustafson, Hannah, University of California, Santa Barbara
Bausch, Andreas, Technical University Munich
Streichan, Sebastian, University of California, Santa Barbara

Related manuscript

The dataset presented here is described in the manuscript "Morphodynamic Atlas for Drosophila Development" by Mitchell et al., https://www.biorxiv.org/content/10.1101/2022.05.26.493584v1. The reader can find details about the motivation, experimental and computer vision methods that underly the dataset.

Funding

This work was supported by:
- National Institutes of Health, Award: R35 GM138203
- National Science Foundation, Award: PHY- 1748958
- National Science Foundation, Award: PHY-1707973

License

This work is licensed under a CC0 1.0 Universal (CC0 1.0) Public Domain Dedication license.

About this README

Author: Nikolas Claussen, nclaussen@ucsb.edu. Dated 11/28/2022.

Description of the data and folder structure

Overview

The data, showing recordings of the surfaces of mostly early Drosophila embryos (stages 5-9), is organized in folders first by genotype of the fly line imaged (e.g. 'WT' for wild type), and then by the fluorophore visualized (e.g. Histone-RFP). Each folder in a genotype/fluorophore combination is indexed by the date in YYYY-MM-DD-HH-MM format, and contains data pertaining to a single experiment. For example, the folder 'toll[RM9]/Bazooka/201908151459' contains data about an embryo from the fly line toll[RM9], stained with an antibody for the protein Bazooka, and imaged on 15/08/2019.

This dataset also contains a set of live recordings of WT embryos at temperatures much higher or lower than standard room temperature conditions (17 and 27 degrees Celsius, respectively). For the purpose of organizing this dataset, these experiments are considered their own "genotype", "WT_17_Degrees" and "WT_27_Degrees".

A list of all datasets is available as a the .csv file 'Morphodynamic_Atlas.csv'. Each row corresponds to a dataset, and the different columns contain the relevant information.

The atlas can be queried (e.g. to find all recordings with a given fluorophore) either using a spread sheet program like Excel, or in Python. A short Python tutorial is provided in 'Morphodynamic_Atlas_Tutorial.ipynb'. The tutorial is also available as a .pdf so you can view it without having Python installed.

The meaning of the columns is explained in what follows, and in the first row of the .csv file:

'Embryo_ID': ID of embryo (time of recording), in YYYYMMDDHHMM format
    'Genotype': Genotype. For organization purposes, recordings at different temperatures are considered genotypes.
    'Fluorofore': Fluorescently tagged protein. For live recordings, we distinguish different tags, e.g. Sqh-GFP and Sqh-mCherry
    'Is_Live': Live or fixed recording
    'Has_PIV': Whether PIV (tissue flow fields) are available. Saved in a PIV/ directory in the same folder as the underlying images
    'Has_PIVLab': Whether high-resolution PIV data (created by PIV lab is available. Saved as XXX_PIVlab.mat file in the same folder as the underlying images
    'Timing_PIV': Timing based on PIV. Indicates frame of movie where PIV switches from ventral furrow to GBE pattern, based on autocorrelation matrix.
    'Timing_Runt': Timing based on Runt stripe 7 deformation. Indicates the corresponding time in Runt reference embryo 202001150004
    'Timing_Warped': Timing based on Runt 7 stripe deformation, after construction of consensus timeline via warping
    'Timing_PMG_VF': Timing based ventral or cephalic furrow for alignment across mutant genotypes
    'Filename': Name of .tif file with image data. Note: some embryos may have multiple .tif files available, e.g. max- and mean projections
    'Time_Resolution_Seconds': Time resolution in seconds
    'Notes': Any additional notes about the dataset

The folders containing the data were zipped into .tar.lz4 archives, with one zipped file for each genotype (e.g. 'toll[RM9]'), with the exception of WT, were, due to the large size of the dataset, each fluorophore has its own zipped file (e.g. 'WT_Sqh-GFP'). The folders were saved as a .tar.lz4 archives, and decompression can be done in a UNIX shell via:

~$ lz4 -d folder_name.tar.lz4 -c | tar xvf -

Tissue cartography conventions

The datasets show the surface of the Drosophila embryo. The Drosophila egg is a curved, ellipsoidal surface, which in this dataset is mapped to the plane by a cylindrical chart in which anterior is left, posterior right, and the ventral side corresponds to the top and bottom edge of the image. See also below, "Timing and cartography metadata".

Multichannel data

Importantly, many recordings in the morphodynamic atlas are multichannel and show two or more different fluorophores. For example, for the WT embryo identifies by timestamp "202001072039", we have a recording of both the proteins "Hairy" and "Runt". The data from the two channels can be found in the respective subfolders /WT/Hairy/202001072039 and /WT/Runt/202001072039. To see whether an embryo of interest was also imaged for other channels, use file search by the embryo time stamp.

Data available for each embryo recording

As explained above, each folder of the form /[genotype]/[fluorophore]/[timestamp] contains data about a single recording.

We now explain what type of data files are available.

Importantly, the dynamic atlas contains both live datasets (i.e. movies of embryos developing inside the microscope), and fixed datasets (embryos killed, fixed, and stained for a particular protein at one moment in development). We first explain what data is available for live recordings:

• .tif files. These are the actual images / movies. Movies are saved as .tif stacks containing multiple frames. Often, more than one .tif file is available. The reason for this is that the movies are z-projections of 3d datasets (as explained in the manuscript linked below), and often, different projections are included. The type of projection is indicated by the file name: "MEAN/mean" indicates a mean of z-slices, "MAX/max" a maximum projection, and "basal" a projection focusing on the basal layer of the cell. Note that for historical reasons, the naming convention for the main .tif files can vary between embryos and genotypes, but the files always show the same information.
• PIV vector fields. For most live datasets, we include a quantification of tissue flow via Particle Image velocimetry. The resulting flow vector fields are saved as .mat file for each timepoint in a separate sub folder, e.g. "PIV". Units of the PIV fields are pixels/frame. Sometimes, more than one version of PIV is included, when different filters where applied (e.g. "PIV_filtered" stores post-processed PIV flow fields in which the velocity field was smoothed), or when a different program was used to compute the PIV (e.g. "PIVlab", using the PIVLab MATLAB plugin). Details on the PIV methodology can be found in the manuscript.
• timing data. For two genotypes, "WT" and "Halo-Hetero Twist[ey53]-Hetero", we used the methodology described in the manuscript to generate timestamps, i.e. for each movie, the frame which corresponds to a common timepoint along the fly embryo morphogenetic trajectory. This is contained in a file called "t0V.txt".
• unit files. All movies in the atlas where taken at a spatial resolution of 0.2619 microns/pixel (though see section below on cartography metadata), and at a time resolution of 1 min?frame, except otherwise noted, but some folders contain files "dt.txt" and "pix2mum.txt", describing the tmeporal resolution in minutes, and the spatial resolution in microns/pixel.
• "info" files. These files are located at the level of the parent folder (i.e. /[genotype]/[fluorofore]) and contain relevant information that applies to the entire set of recordings contained in the folder.
• other files, e.g. "pullbackPathlines", that are of internal nature and where used, for example, to computionally generate the common morphological timeline explained in the manuscript.

The following type of datafiles is available for fixed recordings.

• .tif files. As for live data, these are the actual images, potentially available in different z-projected versions ("MEAN" and "MAX").
• "info" files. These files are located at the level of the parent folder (i.e. /[genotype]/[fluorofore]) and contain relevant information that applies to the entire set of recordings contained in the folder.

For data from the wild type phenotype (WT), additional data is available:

• timing data. WT fixed embryos were time-aligned using the method described in the manuscript. Briefly, all embryos were co-stained with a second marker, Runt, for which live recordings are available. Comparison of a morphological landmark (the shape of Runt stripe 7) to the live data enabled time stamping. For time stamped data, there is a file "timematch_Runt_Runtstripe7_chisq" containing both the time stamp (in minutes, with respect to the live master timeline), and the uncertainty of the timing (in minutes). This file is available as .txt and as .mat. The directory containing the Runt channel of such a fixed embryo contains additional files used by the program that generates the time stamps, notably plots ("stripe7_ssr[...].png", "Runtstripe7_rgb_[...].png") and data on the location and shape of the morphological landmark ("Runt_stripe7curve.mat", "Runt_mask.tif").
• smoothed data. For many WT fixed embryos, per-computed smoothed versions of the image, as well as images showing the x- and y- gradients in fluorescent intensity, are available in sub-folders entitled "sigma050_step001" ect. Here sigma is the standard deviation of the Gaussian used for smoothing, and step the step size used to compute the gradients. This information was computed for use in the paper "Geometric control of Myosin-II orientation during axis elongation" by Lefebvre & Claussen et al., https://www.biorxiv.org/content/10.1101/2022.01.12.476069v1.

Datasets - special cases

In addition to recordings of different fly genotypes during early development, the dataset also includes a movie of a nuclear marker that covers the entire duration of embryo development (WT/H2a-mCherry_Klarsicht/201802181715), and data from opto-genetic experiments (in the genotype folder "opto-RhoGEF2"). In these experiments, the embryo was not simply observed, but manipulated using optogenetic activation, as described in the manuscript "Patterned mechanical feedback establishes a global myosin gradient" by Gustavson et al. (DOI: 10.1038/s41467-022-34518-9).

Timing and cartography metadata

The images and movies contained in the atlas dataset were created from 3-dimensional data using tissue cartography ("Tissue cartography: compressing bio-image data by dimensional reduction", Heemskerk & Streichan 2015, DOI: 10.1038/nmeth.3648). The information defining the map from the cylindrical chart/pullback in which the movies/images of the dataset are displayed and the 3d shape of the fly embryo are contained in the folder "embryo_geometry". Here, there are different .ply 3d meshes of the embryo surfaces, .mat data files and sample MATLAB scripts which can be used for basic tissue cartography operations. Details are described in a separate README contained in the "embryo_geometry" folder.

The folder "timing" contains internal files used by the time-stamping algorithm described in the manuscript.z

Changes made since previously published version: There are three changes to the dataset: 1) added a .csv file listing all recordings available in the dataset, 2) added a python tutorial (jupyter notebook) on how to query the dataset and load data, 3) small renaming changes to fix spelling mistakes in folder names.