Run validation of dandiset metadata using translated LinkML schema

[candleindark]

Use LinkML schema produced in #381 to run validation across manifest files and generate reports.

ssh drogon
git config --global --add safe.directory /mnt/backup/dandi/dandiset-manifests/.git
# exit dragon

# in local machine
git clone drogon.dartmouth.edu:/mnt/backup/dandi/dandiset-manifests
cd dandiset-manifests
for d in 00*; do  linkml-validate -s ../dandischema-linkml-auto-converted/dandischema/models.yaml -C Dandiset $d/draft/dandiset.yaml; done

[yarikoptic]

here is full example on those older dandisets , but you should try on more recent

❯ /bin/ls -d 0000*/0.* | sort -k2 -t/ -r | grep '/0.2[56]' | while read v ; do echo "=== $v"; m=$v/dandiset.yaml; grep schemaVersion $m; linkml-validate -s ../dandischema-linkml-auto-converted/dandischema/models.yaml -C PublishedDandiset $m; done
=== 000054/0.260409.0433
schemaVersion: 0.6.0
dandi namespace is already mapped to Prefix({'prefix_prefix': 'DANDI', 'prefix_reference': 'http://dandiarchive.org/dandiset/'}) - Overriding with mapping to Prefix({'prefix_prefix': 'dandi', 'prefix_reference': 'http://schema.dandiarchive.org/'})
[ERROR] [000054/0.260409.0433/dandiset.yaml/0] Additional properties are not allowed ('@context' was unexpected) in /
[ERROR] [000054/0.260409.0433/dandiset.yaml/0] 'Dandiset' is not one of ['PublishedDandiset'] in /schemaKey
=== 000054/0.260408.2101
schemaVersion: 0.6.0
dandi namespace is already mapped to Prefix({'prefix_prefix': 'DANDI', 'prefix_reference': 'http://dandiarchive.org/dandiset/'}) - Overriding with mapping to Prefix({'prefix_prefix': 'dandi', 'prefix_reference': 'http://schema.dandiarchive.org/'})
[ERROR] [000054/0.260408.2101/dandiset.yaml/0] Additional properties are not allowed ('@context' was unexpected) in /
[ERROR] [000054/0.260408.2101/dandiset.yaml/0] 'Dandiset' is not one of ['PublishedDandiset'] in /schemaKey
=== 000003/0.260218.2052
schemaVersion: 0.6.0
dandi namespace is already mapped to Prefix({'prefix_prefix': 'DANDI', 'prefix_reference': 'http://dandiarchive.org/dandiset/'}) - Overriding with mapping to Prefix({'prefix_prefix': 'dandi', 'prefix_reference': 'http://schema.dandiarchive.org/'})
[ERROR] [000003/0.260218.2052/dandiset.yaml/0] Additional properties are not allowed ('@context' was unexpected) in /
[ERROR] [000003/0.260218.2052/dandiset.yaml/0] 'Dandiset' is not one of ['PublishedDandiset'] in /schemaKey
=== 000070/0.251218.1714
schemaVersion: 0.6.0
dandi namespace is already mapped to Prefix({'prefix_prefix': 'DANDI', 'prefix_reference': 'http://dandiarchive.org/dandiset/'}) - Overriding with mapping to Prefix({'prefix_prefix': 'dandi', 'prefix_reference': 'http://schema.dandiarchive.org/'})
[ERROR] [000070/0.251218.1714/dandiset.yaml/0] Additional properties are not allowed ('@context' was unexpected) in /
[ERROR] [000070/0.251218.1714/dandiset.yaml/0] 'Dandiset' is not one of ['PublishedDandiset'] in /schemaKey
=== 000022/0.251116.2247
schemaVersion: 0.6.0
dandi namespace is already mapped to Prefix({'prefix_prefix': 'DANDI', 'prefix_reference': 'http://dandiarchive.org/dandiset/'}) - Overriding with mapping to Prefix({'prefix_prefix': 'dandi', 'prefix_reference': 'http://schema.dandiarchive.org/'})
^C
Aborted!
=== 000021/0.251116.2246
schemaVersion: 0.6.0
dandi namespace is already mapped to Prefix({'prefix_prefix': 'DANDI', 'prefix_reference': 'http://dandiarchive.org/dandiset/'}) - Overriding with mapping to Prefix({'prefix_prefix': 'dandi', 'prefix_reference': 'http://schema.dandiarchive.org/'})
[ERROR] [000021/0.251116.2246/dandiset.yaml/0] Additional properties are not allowed ('@context' was unexpected) in /
[ERROR] [000021/0.251116.2246/dandiset.yaml/0] 'Dandiset' is not one of ['PublishedDandiset'] in /schemaKey
^C%                                                                                                                                                                                 ❯ 
❯ /bin/ls -d 0000*/draft | while read v ; do echo "=== $v"; m=$v/dandiset.yaml; grep schemaVersion $m; linkml-validate -s ../dandischema-linkml-auto-converted/dandischema/models.yaml -C Dandiset $m; done
=== 000003/draft
schemaVersion: 0.6.0
dandi namespace is already mapped to Prefix({'prefix_prefix': 'DANDI', 'prefix_reference': 'http://dandiarchive.org/dandiset/'}) - Overriding with mapping to Prefix({'prefix_prefix': 'dandi', 'prefix_reference': 'http://schema.dandiarchive.org/'})
[ERROR] [000003/draft/dandiset.yaml/0] Additional properties are not allowed ('@context' was unexpected) in /
=== 000004/draft
schemaVersion: 0.6.0
dandi namespace is already mapped to Prefix({'prefix_prefix': 'DANDI', 'prefix_reference': 'http://dandiarchive.org/dandiset/'}) - Overriding with mapping to Prefix({'prefix_prefix': 'dandi', 'prefix_reference': 'http://schema.dandiarchive.org/'})
[ERROR] [000004/draft/dandiset.yaml/0] Additional properties are not allowed ('@context' was unexpected) in /
[ERROR] [000004/draft/dandiset.yaml/0] Additional properties are not allowed ('includeInCitation' was unexpected) in /contributor/0/affiliation/0
[ERROR] [000004/draft/dandiset.yaml/0] Additional properties are not allowed ('includeInCitation' was unexpected) in /contributor/1/affiliation/0
[ERROR] [000004/draft/dandiset.yaml/0] Additional properties are not allowed ('includeInCitation' was unexpected) in /contributor/2/affiliation/0
[ERROR] [000004/draft/dandiset.yaml/0] Additional properties are not allowed ('includeInCitation' was unexpected) in /contributor/3/affiliation/0
[ERROR] [000004/draft/dandiset.yaml/0] Additional properties are not allowed ('includeInCitation' was unexpected) in /contributor/4/affiliation/0
[ERROR] [000004/draft/dandiset.yaml/0] Additional properties are not allowed ('includeInCitation' was unexpected) in /contributor/5/affiliation/0
[ERROR] [000004/draft/dandiset.yaml/0] Additional properties are not allowed ('includeInCitation' was unexpected) in /contributor/6/affiliation/0
[ERROR] [000004/draft/dandiset.yaml/0] Additional properties are not allowed ('includeInCitation' was unexpected) in /contributor/7/affiliation/0
[ERROR] [000004/draft/dandiset.yaml/0] Additional properties are not allowed ('includeInCitation' was unexpected) in /contributor/8/affiliation/0
[ERROR] [000004/draft/dandiset.yaml/0] Additional properties are not allowed ('includeInCitation' was unexpected) in /contributor/9/affiliation/0
[ERROR] [000004/draft/dandiset.yaml/0] Additional properties are not allowed ('includeInCitation' was unexpected) in /contributor/10/affiliation/0
[ERROR] [000004/draft/dandiset.yaml/0] Additional properties are not allowed ('includeInCitation' was unexpected) in /contributor/11/affiliation/0
[ERROR] [000004/draft/dandiset.yaml/0] Additional properties are not allowed ('includeInCitation' was unexpected) in /contributor/12/affiliation/0
=== 000005/draft
schemaVersion: 0.6.0
dandi namespace is already mapped to Prefix({'prefix_prefix': 'DANDI', 'prefix_reference': 'http://dandiarchive.org/dandiset/'}) - Overriding with mapping to Prefix({'prefix_prefix': 'dandi', 'prefix_reference': 'http://schema.dandiarchive.org/'})
[ERROR] [000005/draft/dandiset.yaml/0] Additional properties are not allowed ('@context' was unexpected) in /
[ERROR] [000005/draft/dandiset.yaml/0] Additional properties are not allowed ('includeInCitation' was unexpected) in /contributor/0/affiliation/0
[ERROR] [000005/draft/dandiset.yaml/0] Additional properties are not allowed ('includeInCitation' was unexpected) in /contributor/1/affiliation/0
[ERROR] [000005/draft/dandiset.yaml/0] Additional properties are not allowed ('includeInCitation' was unexpected) in /contributor/2/affiliation/0
[ERROR] [000005/draft/dandiset.yaml/0] Additional properties are not allowed ('contactPoint', 'includeInCitation', 'roleName' were unexpected) in /contributor/3/affiliation/0
=== 000006/draft
schemaVersion: 0.6.0
^CTraceback (most recent call last):
  File "/home/yoh/proj/dandi/dandischema-linkml-auto-converted/.venv/bin/linkml-validate", line 4, in <module>
    from linkml.validator.cli import cli
  File "/home/yoh/proj/dandi/dandischema-linkml-auto-converted/.venv/lib/python3.13/site-packages/linkml/__init__.py", line 4, in <module>
    from rdflib.plugins.serializers.turtle import TurtleSerializer

[candleindark]

@yarikoptic unable to clone the manifest repo. Some git objects are missing.

At local machine

➜  workshop git clone drogon.dartmouth.edu:/mnt/backup/dandi/dandiset-manifests
Cloning into 'dandiset-manifests'...
remote: Enumerating objects: 66327, done.
remote: Counting objects: 100% (66327/66327), done.
remote: Compressing objects: 100% (50711/50711), done.
remote: fatal: unable to read f7c097994e60c2b58dae464633583b65a6691415
error: git upload-pack: git-pack-objects died with error.
fatal: git upload-pack: aborting due to possible repository corruption on the remote side.
remote: aborting due to possible repository corruption on the remote side.
fatal: early EOF
fatal: fetch-pack: invalid index-pack output

At drogon

ito@drogon:/mnt/backup/dandi/dandiset-manifests$ git cat-file -t f7c097994e60c2b58dae464633583b65a6691415
fatal: git cat-file: could not get object info

[candleindark]

The Additional properties are not allowed errors are related to corrupt data in the archive, #276. I thought we have corrected those data via dandi/dandi-archive#2177.

[candleindark]

It seems that dandi/dandi-archive#2177 indeed fixed the data corruption. The metadata of https://dandiarchive.org/dandiset/000004 doesn't have the corruption and validates against LinkML schema without those errors of Additional properties are not allowed.

The manifest in drogon seems to be quite outdated.

[candleindark]

Heads-up on @context: feeding JSON eliminates the error

Simplest way to silence Additional properties are not allowed ('@context' was unexpected): validate manifests as JSON (or pass --input-format json). linkml-runtime's JSON loader strips JSON-LD reserved keywords (@context, @type, @id, …) before validation; the YAML loader does not. This is appropriate — @context is a JSON-LD keyword, not a model field — but the YAML/JSON asymmetry is easy to miss.

Quick confirmation on live 000004/draft:

• JSON input → No issues found.
• Same data reserialized as YAML via ruamel.yaml → only Additional properties are not allowed ('@context' was unexpected) in /. No includeInCitation/contactPoint/roleName affiliation errors — dandi-archive#2177 appears to have propagated.

Code refs (linkml-runtime main @ 7f98f22)

• Loader.json_clean — docstring "Remove empty values and JSON-LD relics from an input file." Deletes any k.startswith("@") recursively.
• JSONLoader.load_as_dict calls json_clean on every load.
• YAMLLoader.load_as_dict does not.

Implication for the bulk runner: prefer JSON input so @context doesn't drown out real findings.

[candleindark]

Status tally of the four error categories from the sample output

• @context at / — addressed by feeding JSON (per previous comment).
• includeInCitation / contactPoint / roleName on /contributor/N/affiliation/0 — data-side fix from dandi-archive#2177 (closing Add proper migration for "Organization" -> "Affiliation" change #276) has propagated; no affiliation errors on the live 000004/draft.
• 'Dandiset' is not one of ['PublishedDandiset'] in /schemaKey — only remaining tension. Published manifests carry schemaKey: "Dandiset" but validate against PublishedDandiset, whose slot is pinned to "PublishedDandiset". Pydantic tolerates this via the name-prefix hack in DandiBaseModel.ensure_schemakey; LinkML doesn't. Tracked alongside the schemaKey slot-usage work in Handle LinkML migration issue of pydantic2linkml: Impossible to generate slot usage entry for the #389.

[candleindark]

An agent skill has been specified for compiling a validation report across all dandiset metadata, https://github.com/dandi/dandi-schema/tree/linkml-conversion/.claude/skills/dandi-linkml-validation-report.

The report is live at https://github.com/candleindark/dandi-linkml-validation-report/, and will be updated as we improve the LinkML schema.

Relevant problems reveal by the report will be addressed in dedicated issues.