Citation and Publication

To cite this project (datasets, encoding data, methods, modeling, code, website, etc.), please refer to Atanas & Kim et al., 2023 below. Please cite the paper directly and DO NOT cite this repository.

Brain-wide representations of behavior spanning multiple timescales and states in C. elegans

Adam A. Atanas*, Jungsoo Kim*, Ziyu Wang, Eric Bueno, McCoy Becker, Di Kang, Jungyeon Park, Talya S. Kramer, Flossie K. Wan, Saba Baskoylu, Ugur Dag, Elpiniki Kalogeropoulou, Matthew A. Gomes, Cassi Estrem, Netta Cohen, Vikash K. Mansinghka, Steven W. Flavell

* equal Contribution

Links:

• https://doi.org/10.1016/j.cell.2023.07.035
• https://pubmed.ncbi.nlm.nih.gov/37607537/

WormWideWeb

Datasets in this project can be interactively visualized on:

https://wormwideweb.org/

Features include:

• Find specific neurons
• Find specific datasets

Contents

Original files

• fit_results.jld2.bz2: model fit results
• fit_results_lite.jld2.bz2: summarized fit results
• umap_dict.jld2.bz2: UMAP projection data
• analysis_dict.jld2.bz2: analysis-related information
• dict_neuropal_label.jld2.bz2: neuropal label (raw/source data. this maps segmentation rois to labels)
• relative_encoding_strength.jld2.bz2: relative encoding strength posteriors
• deepnet-weights.tar.bz2: deep neural nets weights

Summarized data files using generate_encoding_files() of WormWideWebData.jl (https://github.com/flavell-lab/WormWideWebData.jl)

• neuropal_label.jld2.bz2: matched to roi (use this for analysis)
• neuropal_label.json.bz2: neuropal labels
• fit_ranges.h5.bz2: model fit time segment ranges
• sampled_tau_vals_median.h5.bz2: tau (decay-constants) posterior medians
• relative_encoding_strength_median.h5.bz2: relative encoding strength posterior medians
• tuning_strength.h5.bz2: tuning strength info
• encoding_changes_corrected.h5.bz2: encoding changes info
• neuron_categorization.h5.bz2: encoding categorization info

deepnet-weights.tar.bz2

• contains the trained weights of the neural networks used in this project.
• 3dunet_540nm_voxels: 3D U-Net for segmenting neurons
• head_detector_unet: finding worm head landmark used in ANTSUN registration
• head_detector_unet_0622: an alternative version of the above, optimal for NeuroPAL datasets
• microscope_tracker: detecting keypoints for online tracking on the microscope
• behavior_nir: segmentation of the recorded NIR behavior images for behavior quantification

Neural and behavioral datasets

ANTSUN processed datasets and CePNEM processed model fits and analysis data. Check the project packages and notebooks in the project github repository (https://github.com/flavell-lab/AtanasKim-Cell2023/) on using these datasets.

To get all behavior-neural datasets, please see processed_h5.tar.bz2

processed_h5_scrambled.tar.bz2: contains the scrambled datasets used for the control analysis.

Notes

• list of all datasets: lists the datasets and their metadata and other info
• for head angle-related behaviors such as head angle and angular velocity, θh_pos_is_ventral needs to be used to correct the sign (e.g. dv_correction = θh_pos_is_ventral ? -1 : 1). θh_pos_is_ventral info for each dataset is available in the csv file above ("list of all datasets")

History

• v1: original
• v2: added neural datasets file (processed_h5.tar.bz2)
• v3: added stage location information (processed_h5.tar.bz2)
• v4: adjusted stage location information keys for consistency. added encoding-related and neuropal files for WormWideWeb