search

single-cell-data / TileDB-SOMA

Python and R SOMA APIs using TileDB’s cloud-native format. Ideal for single-cell data at any scale.
https://tiledbsoma.readthedocs.io
MIT License
83 stars 24 forks source link

Visualization of SOMA Experiment #2781

Open danishzmalik opened 1 month ago

danishzmalik commented 1 month ago

I've created a SOMA experiment from around 25 in-house studies, using tiledb_soma's append mode. I am now looking at ways to visualize the data in a similar fashion to how cellxgene is using the census object to visualize gene expression and cell clustering for different studies (See pictures below):

image image

ref: https://cellxgene.cziscience.com/gene-expression , https://cellxgene.cziscience.com/datasets (respectively)

I understand the Census object follows a particular type of schema, but could this be achieved using a single standard SOMA Object, without sending the data over to a public server?

Also, are there any packages/tools I could explore to achieve the same?

johnkerl commented 1 month ago

cc @bkmartinjr @aaronwolen

aaronwolen commented 1 month ago

Thanks for reaching out, @danishzmalik!

As you noted, the visualizations and dashboard capabilities provided by CZI rely on certain schemas and superstructures that aren’t present in standard SOMA experiments, and I'm not aware of packages that achieve the same functionality.

That said, interactive visualizations of SOMA experiments is an area we're very interested in. Feedback like yours is incredibly helpful, so if you have specific elements or features from the CZI dashboard that you find particularly useful or if there are other visualization tools you believe would complement your work with SOMA, please feel free to share!

danishzmalik commented 1 month ago

Hi @aaronwolen , thanks for your response.

We'd be really interested in gene expression visualization (ref: https://cellxgene.cziscience.com/gene-expression ) We would like an interface where we could filter the dataset based on certain metadata fields e.g. publication/study name, tissue, disease, sex etc and select different genes and/or cell types So that we have visualizations rendered showing cell counts and gene expression heat maps for selected genes present in our SOMA experiment. Having a feature like this could really help us with our analysis.

Also, would SOMA experiments be compatible with Rshiny dashboards?