Add trial workflow improvements

[bellerbrock]

Expected Behavior

Requested improvements to the add trial workflow:

• Option to select checks from the supplied accession list rather than supply them as a separate list

• Addition of a "Planting order' field to the trial design, which would be a simple incrementing number that counts up in a serpentine manner. User would indicate the plot where it should start, and whether it should count up and down rows or up and down columns. This is the key identifer when taking planting material to the field and planting the trial.

• Option to organize reps by column rather than by row. For example 12 column trial where each rep is 3 of the columns and each row contains parts of each rep.

• Most fields in workflow stay selected when it's temporarily closed, others like location, design type, and accessions lists do not. would be helpful if they all stayed.

• Other features already documented in other issues: more flexible custom plot naming. Specifically plot numbers based on row and column, not rep and # within rep.

For Bugs:

Environment

Steps to Reproduce

[bellerbrock]

R file that NCSU program has used to generate their trail designs in the past for reference https://drive.google.com/file/d/1rfAqBnHX0HueKQf-ws7NCEXZ-DuNITvl/view?usp=sharing

[lukasmueller]

Is this still an active ticket?

[ch728]

A lot of these workflow improvements are really essential to being able to use the design tool in Breedbase. We are trying to design all of our 50 plus trials in breedbase this year and have come against all of what has been noted here. I would add to this list support for fillers and ability to add borders. Filler example: we have a trial that is 13 accessions replicated three times, which takes 39 plots; however our field is 5 X 8, which means we need to add another accession once as a filler. There is currently no way to do this using the design tool. The borders effect the planting order of accessions, so it would be necessary to have a way of indicating/adding borders. I have built a Shiny App that pulls in the layout via BrAPI and provides some of this functionality to help get our field technicians through the planting season, but it would be ideal to have the functionality directly in BreedBase.

[lukasmueller]

I have been thinking about this - Breedbase allows you to upload physical layout data that overlays on the design. So you can easily enter empty rows in the physical layout. The layout is defined, for each plot in an tab delimited file with plot_name, row_number and col_number. On the relevant trial detail page, open the "Field Layout Tools and Phenotype Heatmap" section and click on "Upload Spatial Layout". If there are row_numbers or col_numbers that are not used, they appear as empty rows (or any combination of missing row_number/col_number will be rendered as an empty cell).

What do you think? Does that solve the problem?

[ch728]

• [ ] I know about the spatial upload, it's handy, but most breeding programs are going to run into the following problems when designing trials: no filler support or support for non-rectangular layouts (5x8 grid with 39 plots), no flexibility in rep orientation, and no concept of planting order. A lot of this can be done if the trial is designed in a spreadsheet beforehand/modified with spatial upload, but having to manually design and manipulate trials in a spreadsheet defeats the purpose of having the database in my opinion. It would be better if a little more flexibility were added to the design tool. Obviously if the user wants to do something very outlandish, then they can fallback to the spatial layout upload/upload existing trial method, but I think there are some small tweaks to the existing breedbase design tool that would cover most layouts for most programs.

[lukasmueller]

Would it be sufficient for the filler support to have the physical layout, and store the metadata of what the filler accession is? (assuming there is only one?). For empty cells, the filler would be returned. Also, planting order seems important, especially when interfacing to automatic planting equipment I suppose. How would you want to specifiy that? Should this be just deduced from the physical rows and columns and some options (zigzag or row by row etc)?

[ch728]

That sounds like it would work for fillers, and yes I think a handful of input parameters + spatial layout would be enough to produce a planting plan. @lukasmueller @chris263 , it might be useful if we could set up a brief meeting at some point. I could show a couple examples of trials that we had a hard time designing in Breedbase/planting plan stuff and then we might be able to 1) find an existing solution that is already implemented in Breedbase or 2) think of some solutions that could be added to make the process easier. On the topic of making Breedbase processes easier, I think it would be useful to have some written, vignette-style tutorials for common tasks, like trial design. This is something I would be willing to help put together.

[lukasmueller]

Great idea to meet. When do you have time? Next week, Monday afternoon looks pretty free for me (after 2pm).

We have an on-line manual that probably needs to be expanded for this purpose. Have you checked it out?

thanks and cheers Lukas

---

From: Chris Hernandez @.> Sent: Friday, April 2, 2021 10:00 AM To: solgenomics/sgn @.> Cc: Lukas A. Mueller @.>; Mention @.> Subject: Re: [solgenomics/sgn] Add trial workflow improvements (#2444)

That sounds like it would work for fillers, and yes I think a handful of input parameters + spatial layout would be enough to produce a planting plan. @lukasmuellerhttps://github.com/lukasmueller @chris263https://github.com/chris263 , it might be useful if we could set up a brief meeting at some point. I could show a couple examples of trials that we had a hard time designing in Breedbase/planting plan stuff and then we might be able to 1) find an existing solution that is already implemented in Breedbase or 2) think of some solutions that could be added to make the process easier. On the topic of making Breedbase processes easier, I think it would be useful to have some written, vignette-style tutorials for common tasks, like trial design. This is something I would be willing to help put together.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/solgenomics/sgn/issues/2444#issuecomment-812543283, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AAAV7FYOKJSZVOFC3S6WEPLTGXEZXANCNFSM4HKXNZ3A.

[ch728]

Haven't checked the docs in a while, it looks like a lot of information has been added since I last checked! There might be some areas that could benefit from expansion and possibly some embedded video examples.

After 2 PM on Monday works for me. How does 2:30 PM sound?

[lukasmueller]

BTW - we also have a youtube channel - search for breedbase in youtube on you'll see it.

cheers Lukas

---

From: Chris Hernandez @.> Sent: Friday, April 2, 2021 11:32 AM To: solgenomics/sgn @.> Cc: Lukas A. Mueller @.>; Mention @.> Subject: Re: [solgenomics/sgn] Add trial workflow improvements (#2444)

Haven't checked the docs in a while, it looks like a lot of information has been added since I last checked! There might be some areas that could benefit from expansion and possibly some embedded video examples.

After 2 PM on Monday works for me. How does 2:30 PM sound?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/solgenomics/sgn/issues/2444#issuecomment-812580459, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AAAV7F3T573QVZTKPJXA26LTGXPSBANCNFSM4HKXNZ3A.

[bellerbrock]

@ch728 Thanks for the additional info! Lukas and I are just looking at this again.

Many of the features discussed are now possible to add through the updated fieldmap viewer (border rows, planting order number, transpose feature to turn row-wise reps into column-wise reps). But those need to be better documented in the workflow, and others like more customizable plot naming are still not implemented.

Let me break those remaining parts out into more manageable individual tickets, then I'll close this one.

[aco46]

@bellerbrock: checking if you were able to create tickets for the remaining tasks in this issue that is yet to be implemented. We could close this and work on individual tickets for remaining tasks.