This app is based on the D3.js data visualization library.
Source code is available at github.com/WormBase/wormcells-viz.

Please note that WormBase reprocesses the original data starting from the author provided gene count matrices.
The data was processed using the scVI model from scvi-tools, a probabilistic framework for single cell RNA sequencing data that explicitly models technical aspects of the data such as batch and sequencing depth. scVI is a generative model that among other things can estimate the gene expression rate of each gene in each cell - what fraction of the transcripts sampled would belong to a gene in that cell.
For details on how the data was processed, please visit the wormcells-viz GitHub repository.

This is a visualization of the Ben-David et al 2019 C. elegans single cell RNA sequencing dataset, which profiled 55,508 L2 larvae cells

Original study:
Whole-organism eQTL mapping at cellular resolution with single-cell sequencing

Eyal Ben-David, James Boocock, Longhua Guo, Stefan Zdraljevic, Joshua S Bloom, Leonid Kruglyak

eLife 2021;10:e65857
https://doi.org/10.7554/eLife.65857

Select cell types and genes to visualize mean gene expression rates.
The dotplots display the same information as the heatmap.

Visualize gene abundances stratified by cell type. Because scVI can estimate the expresssion rate for each gene in each cell while accounting for batch and sequencing depth, we can plot histograms of these gene expression rates for each cell type to glean a bit more information than just with a single number (the mean).

Visualize expression of a gene across all cell types relative to one cell type. Swarm plots are useful for identifying candidate marker genes for a given tissue. On the Y axis they display a set of selected genes, which can be sorted by p-value from differential expression analysis. On the X axis the log fold change of all other cell types is displayed, relative to the chosen tissue. By showing the log fold change of each gene in every tissue, it provides a more complete picture than just a volcano plot for choosing marker candidates.