about summary refs log tree commit diff
path: root/docs/genenetwork_datatypes.md
blob: edf21a936739d91ed8847ad7b16eb193303c05e1 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
# GeneNetwork data types

Notes on the core data types GeneNetwork is built around, compiled from
`factory/docs/gn-gemtext-threads`, `factory/code/genenetwork2`,
`factory/code/genenetwork3`, `factory/code/gn-guile`, and this repo's own
`docs/mariadb_db-webqtl.md` schema dump.

## Species

The `Species` table (`db_webqtl`; full `CREATE TABLE` in
`docs/mariadb_db-webqtl.md`) is the root of GeneNetwork's data hierarchy:
every `InbredSet` (a group/population, e.g. BXD), every `Strain`, and
every chromosome length record (`Chr_Length`) is scoped to a species via
a `SpeciesId` foreign key.

### Fields

Field semantics, from `topics/gn-uploader/adding-species.gmi` (written
while manually adding C. elegans to the database) and cross-checked
against the live schema:

| Field           | Meaning                                                                    |
|-----------------|-----------------------------------------------------------------------------|
| `Id`            | Internal species identifier (auto-increment primary key).                  |
| `SpeciesId`     | Set equal to `Id` after insert; FKs use this, not `Id` (see Known quirks). |
| `SpeciesName`   | A layman's name, e.g. "Mouse", "Roundworm", "Barley".                      |
| `Name`          | Lowercased short name for URLs/API, e.g. "mouse".                          |
| `MenuName`      | UI display name, e.g. "Mouse (Mus musculus, mm10)".                        |
| `FullName`      | Binomial (scientific) name, e.g. "Mus musculus".                           |
| `Family`        | A loose grouping across species, e.g. "Vertebrates", "Nematodes".          |
| `FamilyOrderId` | Sort order within `Family`; derivation undocumented (TODO).                |
| `TaxonomyId`    | NCBI Taxonomy ID, e.g. 10090 (mouse), 6239 (C. elegans).                   |
| `OrderId`       | Sort order for display; same undocumented TODO as above.                   |

### Relationships

- `InbredSet.SpeciesId` -> `Species` -- every group/population (e.g.
  BXD) belongs to exactly one species.
- `Strain.SpeciesId` -> `Species` -- every strain belongs to exactly one
  species (`genenetwork2/gn2/base/species.py`'s `TheSpecies`/
  `Chromosomes` classes join on this).
- `Chr_Length.SpeciesId` -> `Species` -- chromosome names and lengths
  are per species (used by `gn3.db.species.get_chromosome` and GN2's
  `Chromosomes.chromosomes`, both selecting from `Chr_Length` joined to
  either `Species` directly by name, or indirectly via `InbredSet` for
  cases where only a group name is known).
- `GeneIDXRef` cross-references gene IDs between species. GeneNetwork's
  internal gene identifiers are mouse-centric by convention: GN3's
  `gn3.db.species.translate_to_mouse_gene_id` translates a rat or human
  gene ID to its mouse equivalent via this table (returns the ID
  unchanged if the species is already "mouse").
- `InbredSet.SpeciesId` also scopes authorization: GN3's
  `gn3/api/case_attributes.py` resolves a species ID from `InbredSet` to
  build the resource ID used to check a user's access with `gn-auth`
  (https://github.com/genenetwork/gn-auth), so species is part of GN's
  permission hierarchy, not just a data-organization concept.
- In the RDF/SPARQL metadata graph (`topics/next-gen-databases/gn-classification-scheme.gmi`),
  species is one of the three levels of GN's resource classification
  scheme (`gnc:DatasetType`, `gnc:Set`, `gn:Species`), alongside
  `InbredSet` ("Set") and `DatasetType`. Every member of the `Species`
  table has a corresponding `gn:Species`-classified RDF resource;
  `gn-guile`'s `gn/data/species.scm` builds this metadata from SPARQL,
  separately from the MariaDB `Species` table.

### Reference genomes and coordinate generations

Nothing in the schema stores an assembly *identifier*; the only explicit
labels are prose in `Species.MenuName` (verified against `db_webqtl` on
2026-11-11):

| species | labelled assembly |
|---|---|
| human | hg19 |
| mouse | mm10 |
| rat | mRatBN7.2/rn7 |
| fly | dm6 |
| nematode | ce10 |
| arabidopsis | araTha1 |
| medaka | Oryl23 |
| bat | gsor23 |
| turquoise killifish | MPIA_NFZ_2 |
| tomato | GCF_001406875.1 |
| soybean | US DOE JGI-PGF v2 2021 |
| barley, poplar | unlabelled |

Every position column exists in **three generations**: `Mb` (current),
`Mb_2016`, and a legacy mouse `Mb_mm8` -- the same triplication on
`Chr_Length` (`Length`, `Length_2016`, `Length_mm8`), `ProbeSet`
(`Chr`/`Mb`, `Chr_2016`/`Mb_2016`) and `BXDSnpPosition`. The labels above do
not all agree with the numbers:

- **human** is hg19 throughout, in both generations: `Chr_Length` chr1 is
  249 250 621 in `Length` and `Length_2016`, and SOX2 sits at chr3
  181.429-181.431 Mb (hg19 181 429 788; hg38 would be ~181.71).
- **mouse** `Length_2016` chr1 = 197 195 432 (**mm9** exactly), while the
  current `Length` is 197 426 691, which matches no standard build (mm9
  197 195 432, mm10 195 471 971, mm39 195 154 058). The *probes* are
  mm10-era -- Sox2 at chr3 34.47-35.09 Mb (mm10 ~34.65) -- so the label
  mm10 is right for `ProbeSet` while `Chr_Length` is not.
- **rat** current `Length` chr1 = 282 763 074 (**rn6** exactly), i.e. the
  column is the older build although `MenuName` claims rn7
  (mRatBN7.2, chr1 ~260.5 Mb). Rat probe positions are also internally
  inconsistent: Sox2's 44 probes span chr2 1.203-117.592 Mb where rn6 Sox2 is
  ~119 Mb, which reads as unmapped or badly mapped probes rather than one
  build.
- **`Geno`** has a non-null `Mb` for every locus, but `Mb_2016` only for
  human (all 367 403 rows), mouse (23 609), soybean (4 938) and barley (791);
  `Mb_mm8`, a mouse-named column, is also populated for barley (791) -- legacy
  columns reused across species.

Practical consequence: a position taken from this database is "GN's current
coordinate for that species" and should be quoted with that build, not with a
modern assembly. The search index uses the *current* columns
(`ProbeSet.Mb`, `Geno.Mb`), so it inherits exactly these builds -- see
[search-engines.md](search-engines.md#what-the-index-covers-and-whose-coordinates-it-carries).

### Known quirks

- GeneNetwork's genotype tables were originally built mouse-only and
  still carry that legacy: `topics/gn-uploader/genotypes-assemblies-markers-and-genenetwork.gmi`
  notes the `Geno` table has assembly-specific, mouse-named columns
  (`Mb_mm8`, `Chr_mm8`) that don't generalize to other species' genome
  assemblies.
- There is no UI for adding a new species; it's done by hand with
  `INSERT INTO Species(...)` followed by `UPDATE Species SET
  SpeciesId=Id WHERE ...` (see `adding-species.gmi` for the full
  worked example, adding C. elegans).
- `FamilyOrderId` and `OrderId` remain undocumented upstream (flagged
  TODO in `adding-species.gmi`); their exact intended derivation isn't
  recorded anywhere found in these repos.

### Existing implementations

- **GeneNetwork2** (`gn2/wqflask/api/router.py`): `GET
  /api/v_pre1/species` (all species, columns `Id`/`Name`/`FullName`/
  `TaxonomyId`) and `GET /api/v_pre1/species/<species_name>` (one
  species, matched against `Name`, `FullName` or `SpeciesName`, same
  4-column subset). `gn2/base/species.py`'s `TheSpecies` class also
  exposes a species' chromosomes via `Chr_Length`.
- **GeneNetwork3**: `gn3/db/species.py` has `get_all_species`,
  `get_chromosome`, `species_name` (species for an `InbredSet` group)
  and `translate_to_mouse_gene_id`; none of these are exposed as a
  `Species`-table REST endpoint. `gn3/api/metadata.py`'s `GET /species`
  and `GET /species/<name>` are SPARQL-based RDF metadata lookups
  instead, a different data source from the MariaDB `Species` table.
- **gn-guile** (`gn/data/species.scm`): SPARQL/RDF metadata only, same
  category as GN3's `/species` above, not backed by `db_webqtl`.
- **GeneNetwork4** (this repo): `GET /species` (all species, mirrors
  GN3's `get_all_species` query/ordering) and `GET /species/{species}`
  (one species, every column, same match logic as GN2's
  `get_species_info` -- see `README.md`).

## InbredSet

The `InbredSet` table (full `CREATE TABLE` in `docs/mariadb_db-webqtl.md`)
represents a group/population of genetically related individuals within
a species, e.g. BXD (a mouse recombinant inbred panel). Every dataset
(`PublishFreeze`, `GenoFreeze`, `ProbeFreeze`/`ProbeSetFreeze`) belongs
to exactly one `InbredSet`.

**A good alias for `InbredSet` is "Group"** -- this is the term used
consistently across GeneNetwork's own code and docs, not an invention
for this document:

- GeneNetwork2's REST API names the endpoints `GET /api/v_pre1/groups`
  and `GET /api/v_pre1/group/<group_name>` (`gn2/wqflask/api/router.py`),
  mirroring `/species`/`/species/<name>` one level down.
- GeneNetwork2's `DatasetGroup` class (`gn2/base/data_set/datasetgroup.py`)
  docstring: "Each group has multiple datasets; each species has multiple
  groups. For example, Mouse has multiple groups (BXD, BXA, etc)."
- The variable name `group_name` (and `self.group`) appears throughout
  both GeneNetwork2 and GeneNetwork3 wherever an `InbredSet.Name` is
  passed around.
- `topics/documentation/gn-hacking-documentation.gmi` has a "Groups"
  heading stating plainly: "These are in the *InbredSet* table in the
  database."

"RISet" (Recombinant Inbred Set) shows up as a secondary, more
specialist alias -- GeneNetwork2's API docs (`api_readme.md`) call the
same thing "Groups/RISets" and use `group/riset` interchangeably in
prose, and GeneNetwork3's `gn3.db.species.species_name` docstring reads
"given the group (RISet)". "Group" is the more broadly understood term
and the one used in user-facing API routes and UI text; "RISet" is
closer to the underlying genetics (not every `InbredSet` is strictly a
recombinant inbred set -- `GeneticType` also records "intercross"
populations -- so "RISet" is a slight misnomer for the general case).

### Fields

| Field             | Meaning                                                                   |
|-------------------|---------------------------------------------------------------------------|
| `Id`              | Internal group identifier (auto-increment primary key).                   |
| `InbredSetId`     | Set equal to `Id` after insert; FKs (Publish/Geno/ProbeFreeze) use this.  |
| `InbredSetName`   | UI display name -- GN2 calls this `DisplayName` in JSON -- e.g. "BXD".    |
| `Name`            | Short name for URLs/API, e.g. "BXD", "B6D2F2"; usually = `InbredSetName`. |
| `SpeciesId`       | FK to `Species.SpeciesId`; a group belongs to exactly one species.        |
| `FullName`        | Longer descriptive name, e.g. "BXD RI Family", "BXD Family".              |
| `public`          | Numeric visibility level (0-2 seen); threshold semantics undocumented.    |
| `MappingMethodId` | Mapping tool offered, if any; see footnote below the table.               |
| `GeneticType`     | Breeding scheme, e.g. "riset" (as in BXD) or "intercross" (F2/AIL).       |
| `Family`          | Loose grouping across groups, parallel to `Species.Family`.               |
| `FamilyOrder`     | Sort order within `Family` (no `Id` suffix, unlike `Species`).            |
| `MenuOrderId`     | Sort order for display in the group menu.                                 |
| `InbredSetCode`   | Short code, e.g. "BXD", "AXB"; used in precompute/genotype naming.        |
| `Description`     | Free-text description of the group.                                       |

`MappingMethodId` is a key into the `MappingMethod` table (`Id`,
`Name`): 1 `qtlreaper`, 2 `Rqtl`, 3 `HappyR`, 4 `PLINK`, 5 `FastMap`;
null means no mapping tool is offered. This is the authoritative
reading, and the one genenetwork4 uses (`mapping_method` is the
table's name for the id; `GET /mapping-methods` lists the table).
Queried live: 58 groups use 1, 20 use 2, 2 use 3, 22 have none; PLINK
and FastMap are defined but unused. GeneNetwork2's
`get_mapping_methods` (`gn2/base/data_set/datasetgroup.py`) instead
treats the id as a code for a set of tools ("1" = GEMMA + QTLReaper +
R/qtl, "2" = GEMMA only, ...); that reading is superseded and is not
followed here.

### Relationships

- `InbredSet.SpeciesId` -> `Species` -- every group belongs to exactly
  one species (see Species section above).
- `PublishFreeze.InbredSetId`, `GenoFreeze.InbredSetId`,
  `ProbeFreeze.InbredSetId` -> `InbredSet` -- every dataset belongs to
  exactly one group; `genenetwork2/gn2/base/data_set/__init__.py` and
  `datasettype.py` join on these throughout.
- `StrainXRef.InbredSetId` -> `InbredSet` -- links which strains/samples
  are members of a group and whether they're `Used_for_mapping`
  (`topics/systems/mariadb/precompute-mapping-input-data.gmi`).
- `InbredSet.SpeciesId` also scopes authorization the same way as
  described in the Species section: GN3's `gn3/api/case_attributes.py`
  resolves a species ID from `InbredSet` to build the `gn-auth` resource
  ID for a given group.
- In the RDF/SPARQL classification scheme
  (`topics/next-gen-databases/gn-classification-scheme.gmi`), `InbredSet`
  is the "Set" level (`gnc:Set`), the middle of the three-level scheme
  between `gnc:DatasetType` and `gn:Species`.

### Known quirks

- `public` and the `confidentiality` field seen on dataset-level tables
  (`ProbeSetFreeze`, etc.) are separate, differently-scaled visibility
  concepts; neither's exact value semantics were found documented in
  these repos.
- `FamilyOrder` on `InbredSet` has no `Id` suffix, unlike the otherwise
  parallel `Species.FamilyOrderId` -- a naming inconsistency between the
  two tables, not a meaningful difference.
- Not every `InbredSet` is a true recombinant inbred set in the genetics
  sense (see "GeneticType" above and the "RISet" alias discussion) --
  `InbredSet` is GeneNetwork's catch-all table for any group/population,
  regardless of its breeding design.

### Existing implementations

- **GeneNetwork2** (`gn2/wqflask/api/router.py`): `GET
  /api/v_pre1/groups[/<species_name>]` (all groups, optionally scoped to
  a species) and `GET /api/v_pre1/group/<group_name>` (one group,
  matched against `InbredSetName`, `Name` or `FullName`, optionally also
  scoped to a species). `gn2/base/data_set/datasetgroup.py`'s
  `DatasetGroup` class additionally resolves a group's available mapping
  methods from `MappingMethodId`, by its own code table rather than the
  `MappingMethod` table (see the `MappingMethodId` field above).
- **GeneNetwork3**: no `InbredSet`-table REST endpoint or `gn3/db/`
  module was found for groups specifically (unlike `gn3/db/species.py`
  for species).
- **gn-guile** (`gn/data/group.scm`): `get-group-links` and
  `get-group-data`, SPARQL/RDF metadata only, same pattern as
  `gn/data/species.scm`, not backed by `db_webqtl`.
- **GeneNetwork4** (this repo): no `/species/{species}/groups`-style
  endpoint yet.

## PublishXRef

The `PublishXRef` table (full `CREATE TABLE` in `docs/mariadb_db-webqtl.md`)
is the central record for a published (phenotype) trait within one
group: one row per trait per group, tying together that trait's
`Phenotype` description, its `Publication` reference, its actual
per-strain values (`PublishData`), and -- for traits that have been
mapped -- a single best QTL hit (`Locus`/`LRS`/`additive`/`mean`).
Explored at length in `topics/systems/mariadb/precompute-publishdata.gmi`,
which walks through precomputing GEMMA hits for every `PublishXRef` row
and surfacing its `Phenotype`/`Publication` metadata alongside them.

### Fields

| Field           | Meaning                                                                        |
|-----------------|--------------------------------------------------------------------------------|
| `Id`            | Trait id within a group; unique only paired with `InbredSetId` (Known quirks). |
| `InbredSetId`   | FK to `InbredSet`; every trait belongs to exactly one group.                   |
| `PhenotypeId`   | FK to `Phenotype`, the trait's description and units.                          |
| `PublicationId` | FK to `Publication`, the article the trait was published in.                   |
| `DataId`        | FK to `PublishData.Id`, the trait's actual per-strain values.                  |
| `mean`          | Mean value across strains, for the trait's single stored QTL hit.              |
| `Locus`         | Marker/SNP name of that stored hit.                                            |
| `LRS`           | LOD/LRS score of that hit.                                                     |
| `additive`      | Additive effect size of that hit.                                              |
| `Sequence`      | Disambiguates rows colliding on `(InbredSetId, PhenotypeId, PublicationId)`.   |
| `comments`      | Free-text comment on the mapping.                                              |
| `NSamples`      | Number of strains/samples the trait was computed over.                         |

### Relationships

- `PublishXRef.InbredSetId` -> `InbredSet` -- every trait belongs to
  exactly one group (see InbredSet section above).
- `PublishXRef.PhenotypeId` -> `Phenotype` -- the trait's
  `Pre_publication_description`/`Post_publication_description`/
  `Original_description` and `Units`.
- `PublishXRef.PublicationId` -> `Publication` -- the article the trait
  was published in (`Title`, `Authors`, `Journal`, `Volume`, `Pages`,
  `Month`, `Year`, `PubMed_ID`, `Abstract`).
- `PublishXRef.DataId` -> `PublishData.Id` -- the trait's actual
  per-strain values; `PublishData.StrainId` -> `Strain` names them.
  `PublishSE` (per-strain standard error) and `NStrain` (per-strain
  sample count) key off the same `(DataId, StrainId)` pair, both
  optional (left-joined in GeneNetwork2's own queries).

### Known quirks

- `PublishXRef.Id` is not globally unique on its own -- only the pair
  `(InbredSetId, Id)` is (`UNIQUE KEY InbredSet`) -- so a group must
  always be known to look up a trait by Id. Observed live: Id `10001`
  exists under dozens of different groups, each with a different
  `DataId`.
- `Publication` coverage is spotty: per precompute-publishdata.gmi's
  own count, about 22K of 29K `Publication` rows lack an `Abstract`,
  and many `PubMed_ID` values look implausible for a real PubMed ID
  (too few digits to be one). Treat `Publication` metadata as
  best-effort, not authoritative -- the doc's own conclusion is "It
  really is a bit of a mess."
- Only a single highest-LRS hit is stored per trait
  (`mean`/`Locus`/`LRS`/`additive`); precompute-publishdata.gmi and
  precompute-mapping-input-data.gmi are both about extending this to
  store more hits (e.g. every GEMMA SNP above a threshold, not just
  the best one) -- not yet reflected in the schema.

### Existing implementations

- **GeneNetwork2** (`gn2/wqflask/api/router.py`): `trait_sample_data`'s
  Publish branch returns per-strain `PublishData` values (joined with
  `PublishSE`/`NStrain`); `get_trait_info`'s Publish branch returns
  only `PhenotypeId`/`Locus`/`LRS`/`additive`, without the
  `Phenotype`/`Publication` join.
- **gn-guile** (`gn/data/dataset.scm`): `get-bxd-publish-list`,
  `get-bxd-publish-values-list`, `get-bxd-publish-name-value-dict` --
  hardcoded to `InbredSetId=1` (BXD) throughout, no `Phenotype`/
  `Publication` join.
- **GeneNetwork4** (this repo): `GET /groups/{group}/traits/{trait_id}`
  (the `PublishXRef`/`Phenotype`/`Publication` join above, generalized
  to any group) and `GET /groups/{group}/traits/{trait_id}/values`
  (per-strain `PublishData` values).

## ProbeSetXRef

The `ProbeSetXRef` table (full `CREATE TABLE` in
`docs/mariadb_db-webqtl.md`) is the expression (mRNA) counterpart to
`PublishXRef`: one row per probe/gene per expression dataset, tying
together a probe's actual per-strain values (`ProbeSetData`) with a
single best QTL hit (`Locus`/`LRS`/`pValue`/`additive`/`mean`). Unlike
`PublishXRef`, it has no single `Id` column of its own -- its primary
key is `DataId`, and the natural key identifying one row is the pair
`(ProbeSetFreezeId, ProbeSetId)`.

### Fields

| Field              | Meaning                                                             |
|--------------------|---------------------------------------------------------------------|
| `ProbeSetFreezeId` | FK to `ProbeSetFreeze`; the expression dataset this row belongs to. |
| `ProbeSetId`       | FK to `ProbeSet`; the probe/gene this row's mapping belongs to.     |
| `DataId`           | Primary key; FK to `ProbeSetData.Id`, the per-strain values.        |
| `mean`             | Mean expression value across strains.                               |
| `se`               | Standard error of the mean.                                         |
| `Locus`            | Marker/SNP name of this trait's single stored QTL hit.              |
| `LRS`              | LOD/LRS score of that hit.                                          |
| `pValue`           | p-value of that hit.                                                |
| `additive`         | Additive effect size of that hit.                                   |
| `h2`               | Heritability estimate.                                              |

### Relationships

- `ProbeSetXRef.ProbeSetFreezeId` -> `ProbeSetFreeze` -- the expression
  dataset (e.g. a tissue/platform/batch combination) this mapping
  belongs to. `ProbeSetFreeze.ProbeFreezeId` -> `ProbeFreeze.ProbeFreezeId`
  (the same "set equal to the primary key after insert" pattern as
  `Species.SpeciesId`/`InbredSet.InbredSetId`) -> `ProbeFreeze.InbredSetId`
  -> `InbredSet` is how an expression dataset is scoped to a group;
  `ProbeFreeze.TissueId` -> `Tissue` names the tissue it was measured
  in.
- `ProbeSetXRef.ProbeSetId` -> `ProbeSet` -- the probe/gene annotation
  (`Name`, `Symbol`, `description`, `Chr`, `Mb`, `alias`, `GeneId`, ...).
  `ProbeSet.Name` is only unique per `(ChipId, Name)`, not globally, but
  is unique in practice within one dataset's member probes, which is
  how GeneNetwork2 and this repo both address a probe (dataset name
  plus `ProbeSet.Name`) without needing `ProbeSetId` in the URL.
- `ProbeSetXRef.DataId` -> `ProbeSetData.Id` -- the trait's actual
  per-strain values; `ProbeSetData.StrainId` -> `Strain` names them.
  `ProbeSetSE` (per-strain standard error) keys off the same
  `(DataId, StrainId)` pair, optional (left-joined in GeneNetwork2's
  own queries), same pattern as `PublishSE` for `PublishData`.

### Known quirks

- No `(ProbeSetFreezeId, ProbeSetId)` `UNIQUE KEY` is declared in the
  schema (only separate, non-unique `KEY`s on each column) -- a probe
  is assumed to appear at most once per dataset by convention, not by
  a database constraint; both GeneNetwork2 and this repo rely on that
  assumption (`exec_first`/`fetchone()`).
- Only a single highest-LRS hit is stored per probe, the same
  limitation `PublishXRef` has (see its Known quirks above) -- the
  same precompute docs discuss extending both tables to store more
  hits.

### Existing implementations

- **GeneNetwork2** (`gn2/wqflask/api/router.py`): `get_datasets_for_group`
  (list, `ProbeSetFreeze` joined to `ProbeFreeze`/`InbredSet`),
  `get_dataset_info`'s ProbeSet branch (one dataset, joined to
  `Tissue`), `trait_sample_data`'s ProbeSet branch (per-strain
  `ProbeSetData` values), `get_trait_info`'s ProbeSet branch
  (`ProbeSet` annotation plus `ProbeSetXRef` mapping stats, minus
  `pValue`).
- **GeneNetwork4** (this repo): `GET /groups/{group}/datasets` (list),
  `GET /datasets/{dataset}` (one dataset -- not scoped under
  `/groups/{group}`, since a dataset name is globally unique on its
  own), `GET /datasets/{dataset}/probesets/{probeset}` (the
  `ProbeSetXRef`/`ProbeSet` join above) and
  `GET /datasets/{dataset}/probesets/{probeset}/values` (per-strain
  `ProbeSetData` values).