SeqFish demos - Githubissues

haniffalab / webatlas-pipeline

A data pipeline built in Nextflow to process spatial and single-cell experiment data for visualisation in WebAtlas

MIT License

46 stars 10 forks source link

SeqFish demos #80

Closed davehorsfall closed 1 year ago

davehorsfall commented 1 year ago

There are several issues with the integrated SeqFish data demos that need to be discussed.

The performance is slow. Selecting genes causes the app to hang for several seconds.
The seqFISH dataset alone is showing tile positions rather than cell types (if @BioinfoTongLI does not have cell types for seqFISh alone, either use the annotation from the integrated object or remove the individual seqFISH and scRNAseq modalities of mouse atlas)

BioinfoTongLI commented 1 year ago

re 1. the slowness might comes from the sc data not being chunked (?). re 2. the celltype information is in obs/celltype_mapped_refined and the working example is here. https://webatlas.cog.sanger.ac.uk/latest/index.html?config=https://bayraktar.cog.sanger.ac.uk/Seqfish/0.0.1/seqFish-embryo1-config.json

dannda commented 1 year ago

Couldn't reproduce the issue when testing on a local Vitessce instance (from commit 33c30b5b2b6088ec8cd404fe76c6d56b15d6aa42). Loading and rendering had no problems.

BioinfoTongLI commented 1 year ago

@dannda do you mean the isssue is only on webatals hosted on S3 buckets (non-local version) ?

dannda commented 1 year ago

Oh no, I just think there was some later development that might have fixed that performance issue after the current Vitessce version we have hosted. But still unsure where exactly the issue happens so would have to investigate more; I just noticed when I tried to reproduce the issue locally but with Vitessce's main branch

BioinfoTongLI commented 1 year ago

this seems to be resolved by bumping vitessce version to v3? If so, please feel free to close this @davehorsfall @dannda

davehorsfall commented 1 year ago

Fixed by bumping webatlas-app to vitessce 3.2.2.