Repository navigation
Conversation
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
![claude[bot]](https://avatars.githubusercontent.com/in/1236702?s=60&v=4){kind=small}
|
@claude review |
| evaluation: | ||
| mode: diagnostic | ||
| accepted_outcomes: | ||
| - outcome: refused_for_missing_values |
There was a problem hiding this comment.
Is this part of the protocol standard? Probably what we needed is "error" or "refused"
There was a problem hiding this comment.
The standard allows any label, and its own example uses refused_for_orientation. If you want a fixed set of labels, that would be a question for the standard repo.
|
Hi @ohih8 , please see waldronlab/agent-protocol-standard#46 for instructions on clarifying how to specify the rarefaction recommendation in this PR. |
Move every input condition into Materials as must/should conditions (ADR 0017): a non-rarefied count table now proceeds with a warning that advises rarefaction, and the separate validation step is removed. Consolidate the rarefaction text, drop depth normalization as a material, name the vegan and scikit-bio calls, and loosen the proportion-sum tolerance to 1e-6. counts_and_zeros becomes a diagnostic test that requires the warning alongside the results; its expected values are confirmed with vegan and scikit-bio. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…versity # Conflicts: # PROTOCOLS.yaml
|
still need to think about this: Come up with a strict definition of what belongs in materials, and how materials are validated. |
I've done some formalizing of what belongs in materials in ADR 0017, Conditions on Materials, and updated the standard accordingly (waldronlab/agent-protocol-standard#47). What belongs in Materials
Two tiers for Materials requirements
I've also added agent skills there to create, update, and validate protocols, which will help enforce these rules. Welcoming feedback if this leaves any ambiguity or problems! |
|
Hi Dr. Waldron, over the last two days Ive updated the PR, and resolved the comments (many were already fixed from the standard changes). I left one up that asks about the error name, but it doesn't affect the protocol itself much. |
lwaldron
left a comment
lwaldron
left a comment
There was a problem hiding this comment.
Looks great - one minor fix not to specify a file format when unnecessary (that I should add to guidelines) then I'll merge.
|
|
||
| ### Step 3: Write the result | ||
|
|
||
| Write a CSV table with exactly two columns, `sample_id` and `shannon`, and one row |
There was a problem hiding this comment.
Doesn't have to be csv - "create a table" should be enough. You might not want to write it to disk before using it in your next step of analysis.
The input and result are tables, not necessarily CSV files, so the protocol no longer names a format or implies writing to disk. Adds the approved review of 1.0.0 to the frontmatter and History. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
Fixed the file format stuff and added the review, the one thing is that the Notes section currently reads:
This is just a placeholder, so if you could add your own thoughts that would be great. The standard requires that the Notes section not be empty, I think, but I don't just want to put words in your mouth for the review. Otto |
…n tests this week. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
… hand" with a calculator. Note the only bad-data test that accepts something other than refusal is the matrix transposition one, because the protocol specifically states to not fix other things (like renormalizing, imputing, etc.). More tests could probably be added but these are a few to start.
Closes #50
Adds Shannon diversity protocol and two small tests for it. More tests should be added in the future, however these are the most important and it is probably a good idea to wait until the testing infrastructure is fully set up before generating a large amount of tests.
NOTE: This Protocol does not include rarefaction or sequencing depth-normalization as an intentional design choice by me to simplify the protocol. It can certainly be added if desired, however I wanted to keep the protocol concise.