[Placeholder / Discussion] Draw a picture of what the QC report will contain for MVP

[jaclyn-taroni]

A good way to make sure we're all on the same page about the contents and layout of the QC report is to sketch out what the overall report will look like.

From there, we can 1. disagree a little bit initially, probably 2. ensure we have issues for all the relevant sections 🚀

[allyhawkins]

In an effort to get this started I went ahead and created a very very rough drawing of what I think we are looking for in the QC report. The first two tables, sample_information and processing_information we already have a version added to the report so I just filled in some examples of data that is in there but it is not complete so as not to crowd the picture.

I also tried my best to draw out the plots but in general, I think we would have the following plots:

• our usual plot we show during training with UMI/cell on the x-axis, genes/cell on the y-axis and colored by mito content
• knee plot where we are coloring cells based on if they are included in the filtered sce object or not (filtering here referring to emptyDrops filtering)
• plot showing the miQC model with unique genes on the x-axis and mito on the y-axis and showing the model. Would we rather change this and maybe color the cells based on if they are keep or throw based on miQC?

I also added in a section for ADT/ CITE-seq information and noticed on the issue that there was a comment to add in the list of antibodies. I wasn't sure how lengthy that could get or how we would want to portray that so I thought we might want to have that separate from the statistics?

I just wanted to get a general idea of the order of things and the blocks that we need to put in place so this is pretty rough but would love to hear @jaclyn-taroni and @jashapiro thoughts/ opinions/ disagreements/ suggestions. I also wanted to keep in mind that this is for MVP so I did not choose to include any UMAPs or clustering at this point.

[jaclyn-taroni]

This seems perfectly reasonable to me for MVP, which is to say I have no disagreements. I'm uploading a version of that sketch with lettered panels just in the interest of making the next steps easier to talk about.

If @jashapiro also thinks this seems reasonable, I think the create/update issues part of this is as follows (I'm working from inst/rmd/qc_report.rmd):

• Identify a way to make a "two-column" report. If there is no reasonable solution to this problem, revisit the organization using a single column.
• Ensure A ("Raw Sample Metrics") has all the information we need already. If not, file an issue with what needs to be added/updated.
• Ensure B ("Pre-processing Information") has all the information we need already. If not, file an issue with what needs to be added/updated.
• Update Attach basic statistics of per cell QC metrics in QC report #18 to reflect C
• Update Add plot to QC: Total count vs. gene counts by mito content #39 to reflect D
• Ensure E ("Knee plot") is done. If not, file an issue with what needs to be added/updated. (Can we close Add knee plot to QC report #19?)
• Create an issue for F
• Update and/or split up Add a CITE-seq section to the QC report #38 to reflect G & H

Issues should have enough implementation details to get folks headed in the right direction. If we make them more specific than this sketch, which we should aim for, we can have any disagreements about details there.

[jashapiro]

I looked into the two column thing: Looks like we can do that with "fenced divs" https://bookdown.org/yihui/rmarkdown-cookbook/multi-column.html

Basically we add a bunch of ::: {} elements where we want to create divs, and enclose ones we want side by side with ::::{style="display: flex;"}

WE can try it at least!

[jaclyn-taroni]

👍🏻 new issue please :)

[jashapiro]

I do wonder if we should combine A&C? Or move the number of cells to C? Conceptually and statistically, the number of cells after filtering and UMI/cell are related, so it seems like we might want to keep those together.

[allyhawkins]

Right now we have the following included in A:

• Sample ID
• Tech version
• Number of reads sequenced
• Number of mapped reads
• Number of cells reported by alevin-fry

The statistics mentioned in #18 that would be part of the table in C include median UMI/cell, median genes detected/cell, and median mito %/cell. We could append that to what we have currently in A? or We could take Number of cells reported by alevin-fry and move that to C and keep A as a separate table with information about the whole sample rather than at a cellular level?

[jashapiro]

Ah, okay.. I was misunderstanding the # of cells as the number after filtering... but it is only after the barcode filtering, not emptyDrops. I think that is fine to leave in A.

We should add # of cells post-emptyDrops to #18, and I will add that there.

[allyhawkins]

We should add # of cells post-emptyDrops to #18, and I will add that there.

Agree with this, I also forgot that we included the pre empty-drops number of cells when we added that table for A to the QC report originally. Sorry about the confusion.

[jashapiro]

I believe we can close this, as I think we have all content we expect for MVP, with maybe a new issue for "make it pretty" later?

[jaclyn-taroni]

I believe we can close this, as I think we have all content we expect for MVP, with maybe a new issue for "make it pretty" later?

Yup, agreed - go for it!

[jashapiro]

Closing for future work.