DPL-066 Cardinal - Upload cell count data

[KatyTaylor]

User story As a user of the Cardinal pipeline, I want there to be a record of the cell count and other measurements of each sample in the LCA PBMC plate, and to have some help in deciding which wells to take forward and which to fail, based on quality measures.

Who are the primary contacts for this story Stef L

Acceptance criteria

• I can upload a new quant type in quanthub: Cardinal_PBMC_Count
• I can upload an excel (CSV) file to quanthub. See example below. Well name will display the name of each sample as it was in the LIMS-generated Cellaca upload file (rather than A1, A2 etc). For most plates, multiple files will need to be uploaded as only 24 samples can be read at a time on the Cellaca. Quanthub will need to assemble these multiple uploads to be processed together as a single plate. Figure 2. Cellaca output file to be uploaded to quanthub. Up to 4 files per 96 well plate will be generated.

Example files (choice of 2 file formats): Cellaca Excel.xlsx Cellaca Export.csv

• If the % viability is below 50% the well needs to be flagged, to inform the user of lower than optimal viability.
• If live cell count is less than 400,000 cells/ml the well needs to be flagged, to inform the user of lower than optimal cell count.
• Investigate how these flags are normally created - assume Quanthub will both visually highlight them, and export the info to MLWH, and Limber will read this info and also visually highlight them. (new Limber plate presenter, importing required infor from SS, then specifically show this data in Limber)
• I can fail certain wells of choice on the LCA PBMC plate. (should be covered by standard Limber functionality) (Q: assume this will be a manual step, choose reason why failed, which trigger event to MLWH, add more 'reasons', or could be automated with 'threshold' reason)
• I can download the collated quality information so that it can be uploaded to an ELN - if this is not trivial, it's not necessary as they can upload the raw output from Cellaca into ELN.

Dependencies This story is blocked by the following dependencies:

• pipeline setup - https://github.com/sanger/limber/issues/772
• Decide for sure what to put as the sample identifier in the generated input file - https://github.com/sanger/limber/issues/757 (probably sample name = tube barcode)
• LCA PBMC plate created, in https://github.com/sanger/limber/issues/757

[KatyTaylor]

Q's answered by Stef 08/07/21:

1. The flags for <50% viability and <400k live cell count - do these need to be recorded (e.g in MLWH) (probably default behaviour) or just shown on screen for immediate decisions to be made? - would be good if this info is available in Limber for failing wells, or in MLWH for reports.
2. Failing wells - expecting this to be done in Limber, or Quanthub? - think Limber.
3. "Enable the collated quanthub document to be downloaded" - or exported to SS / Limber? - think they want this file for the ELN - if it's more work they could just save the 4 separate files. (not known which ELN they're using or what file format it expects)

[andrewsparkes]

Q. Are the thresholds likely to change over time? Influences whether we need to have them accessible in a table or fixed in config. If changeable do we need change history recorded?

[KatyTaylor]

Q. Are the thresholds likely to change over time? Influences whether we need to have them accessible in a table or fixed in config. If changeable do we need change history recorded?

Good question. There are other similar thresholds in Quanthub, and I think that they are not changeable by the customer, but they are easy enough for us to change, and I also remember there being versions of quant types, so we could always make a new version with a new threshold if the old one is also still required.

[andrewsparkes]

I think I have some thresholds in the purposes (or pipeline) configs in Limber. Requires a re-deployment to change them but easy to find.

[KatyTaylor]

Info from Slack:

James 2nd Sep:

Story DPL-065 (https://github.com/sanger/limber/issues/757) says The LIMS needs to populate the sample name into the respective field (instead of the ‘A1’, ‘A2’ etc shown). However, the resultant output file in DPL-066 (https://github.com/sanger/quanthub/issues/117) shows the well in the well name column, and I was lead to believe that this came from the input file we generated. Given we use the first column to map back to the matching well for recording the data, well name is much more reliable, as it will never be duplicated, and is generally speaking just much easier for us to grab. What was the driver behind listing sample name in these cells?

Stef:

Hi James, i think just having the well ID (instead of sample name) is an option as long as we can attach the plate barcode to the exported cellaca file somehow. Is it enough for quanthub if we put the plate barcode in the title of the output file or would we need to have within the output file itself somewhere? @Stephen Watt what is the best way to include the plate barcode in the cellaca output file? Is there a field where this could be included?

James:

We opted for direct upload via limber, so you’d navigate to the plate itself to attach the QC file, rather than sniffing barcodes or file content. But I understand the concerns with making sure the right file gets uploaded. We could name the cells “plate_barcode:well_name” which complicates @steve’s story slightly, but may be the most robust way of persisting that information.

Stef:

yes that sounds like a nice solution!

[KatyTaylor]

Closing, as is in Done column.