Add celltyping to external instructions #502
Conversation
sjspielman
requested review from
allyhawkins and
jashapiro
and removed request for
allyhawkins
|
I wonder if we want to wait a bit on this PR until we have implemented #470? The reason is that we will be able to better model "good" file names and version tracking at that point. |
Yeah, that seems fine! I think we can leave this PR open and I can come back to it later. I don't foresee many, if any, file conflicts. |
|
Time to revive this PR! I struggled a little bit with the level of detail to provide here vs provide in #555 (updates to internal instructions; please have a peek there for context!). Related to what level of detail we want to achieve here - this comment from before remains unaddressed, in part because we changed PanglaoDB references to be
|
sjspielman
requested review from
allyhawkins
and removed request for
allyhawkins
|
Ready for another look! Again, let's aim to get this in soon since this PR is getting too big, and we can follow up with a subsequent PR where necessary. Note that I left some |
| The `feature_barcode_file` itself is a tab separated file with one line per barcode and no header. | ||
| The first column will contain the barcode or antibody ID and the second column the barcode nucleotide sequence. | ||
| For example: | ||
| By default, `SingleR` annotation uses references from the [`celldex` package](https://bioconductor.org/packages/release/data/experiment/html/celldex.html). |
There was a problem hiding this comment.
Can you file an issue to update this section on cell type references, after #565 goes in.
There was a problem hiding this comment.
fyi im on this, have a note waiting for it :)
| --celltype_project_metafile examples/example_project_celltype_metadata.tsv | ||
| ``` | ||
|
|
||
| ### Providing existing cell type labels |
There was a problem hiding this comment.
No I did mean to move it down here, but also that we could tackle it separately. I think what you have now is fine though!
There was a problem hiding this comment.
The external instructions look good, but I do want https://github.com/AlexsLemonade/scpca-nf/pull/502/files#r1394368472 completed before we merge this in (or an issue about those changes, but it seems small enough to sneak in here).
|
Actually updated the What do you think? |
examples/README.md
Outdated
| - An example [Nextflow configuration file (`user_template.config`)](../external-instructions.md#configuration-files) for the `scpca-nf` workflow. | ||
| - An example [`run_metadata.tsv` file](../external-instructions.md#prepare-the-run-metadata-file) for the `scpca-nf` workflow. | ||
| - An example [`sample_metadata.tsv` file](../external-instructions.md#prepare-the-sample-metadata-file) for the `scpca-nf` workflow. | ||
| - An example [`multiplex_pools.tsv` file](../external-instructions.md#multiplexed-cellhash-libraries) for the `scpca-nf` workflow. | ||
| - An example [`project_celltype_metadata.tsv` file](../external-instructions.md#preparing-the-cell-type-project-metadata-file) for performing optional cell type annotation in the `scpca-nf` workflow. | ||
| - An example [submitter cell type annotation file](../external-instructions.md#providing-existing-cell-type-labels) for optionally providing previously-obtained cell type annotations to the `scpca-nf` workflow. |
There was a problem hiding this comment.
I know I said maybe the file names, but if we do that, we should use the actual file names in this directory. And because of that, I'm rethinking that idea... maybe just the description is sufficient for people to infer the name of the file?
Oh, I missed that comment. I like this version! |
…e text to say run metadata file
…exsLemonade/scpca-nf into sjspielman/499-external-celltype-docs
|
Updated, but in table form which gives better spacing, now that we're using colons. I also removed some changes that went in but in retrospect I think were overkill, and also we decided not to do them - #469 (comment) There is no longer an example submitter file. I think if we circle back later and decide we want it included in the example output, it should be a separate issue/PR discussion. But, there is still a section that says you could provide a submitter file if you want. |
|
@allyhawkins tagging you back in too if you have more thoughts on submitter example files..sorry for all the back and forth on that!!! |
Look good to me, I don't have any other suggestions on that front. |
|
👍 |
Closes #499
This PR adds external cell type documentation. There is a decent amount here, so I expect >1 rounds of review! (reminder that I am out this Thursday+Friday, which may affect when reviewers want to schedule their time to review this!).
I have added documentation throughout for adding submitter annotations and for performing cell typing, including an example celltype metadata file. Any organizational or wording feedback let me know! There is also one
TODOin there about whether we want to say "want a different organ for cellassign? let us know," so please make sure to comment there.Worth noting that the internal docs are still under review (#500) and this PR is not stacked; we may want to add more relative links between these docs files, but I'd like to wait until #500 is all set first. Any more relative links can be added later in this PR or in a subsequent PR.
While working on these docs, I had some thoughts about how we have set up the cell type annotation metadata file - specifically, do actually want this file to include the columns
singler_ref_fileandcellassign_ref_file? You'll see the docs are a bit weird (although I think fairly clear) for the content of these columns. We may wish to alter the cell type process to only accept singler & cell assign reference names, and we would create the meta variables for the reference paths themselves in the celltype annotation subworkflow. If we want to take that approach, I'd update docs here accordingly and then in a separate PR we can update how we handle the metadata in the workflow.