Add basic CELLxGENE workflow #75
Conversation
|
|
||
| # Analysis | ||
|
|
||
| - Follow the Seurat tutorial to calculate marker genes |
There was a problem hiding this comment.
This will take too much time. I'd not do something as comprehensive as this.
There was a problem hiding this comment.
We already have cell types so it should just be one function call to get the markers but happy to do something else. We could also have it as code that isn't run but then there will be no results to save.
There was a problem hiding this comment.
Ok! What you suggest sounds good!
| # Analysis | ||
|
|
||
| - Follow the Seurat tutorial to calculate marker genes | ||
| - Save results as a text file |
There was a problem hiding this comment.
Can we not use a text file but simply save an updated seurat object or alternatively a dataframe/parquet?
There was a problem hiding this comment.
Data frame should be possible (I wrote this before I reminded myself how saving works). Saving a Seurat would probably require converting back to an AnnData if we are doing things through {reticulate}.
There was a problem hiding this comment.
Saving a Seurat would probably require converting back to an AnnData
Not necessarily. It's perfectly legit if an R team works with their Seurat objects internally as long as they don't expect an ML team to train on these artifacts.
|
|
||
| # Add results | ||
|
|
||
| - Add the results as an artifact |
There was a problem hiding this comment.
This is part of the previous step.
| # Add results | ||
|
|
||
| - Add the results as an artifact | ||
| - Render and add report (?) |
There was a problem hiding this comment.
You need to tell me but can we something like the following?
lamin save myanalysis.qmdMeaning that you have a markdown-like file with everything in it.
The following would already work if the notebook contained Python code. I can make this work for a notebook that contains R but you should tell me whether that's necessary if we rather go with R markdown.
lamin save myanalysis.ipynbThere was a problem hiding this comment.
We could use a .ipynb to run code but I'm not sure we could add it to the current website (if that's the place). Also, it's not commonly used for R analysis. There is also Quarto which can be used for R and Python.
I'm not sure if we could run lamin save inside the same document or if that would break. I don't know enough about how it all works to say what the best approach might be.
There was a problem hiding this comment.
Also, it's not commonly used for R analysis.
Then this should be the last resort. We should use what's commonly used.
I don't know enough about how it all works to say what the best approach might be.
We should think about the best approach for an R user, not about what's easiest for lamin. I'll try to call you.
There was a problem hiding this comment.
@lazappi and I just went on a call and hashed out the UX.
- The user exports their current
.qmdor.rmdsession as an html with the same filename but the.htmlsuffix - The user runs
lamin save myanalysis.qmdon the terminal
If the user forgets to export a .html the CLI will raise an error and remind the user to do that manually.
| ) | ||
| ``` | ||
|
|
||
| # Introduction |
| - Save and upload results | ||
| - Save and upload report (?) | ||
|
|
||
| # Connect to a LaminDB instance |
| - Connect to CELLxGENE instance | ||
| - Show instance object | ||
|
|
||
| # Downloading a dataset |
Add a vignette showing basic usage of the CELLxGENE instance