Add barseq acquisition by dougollerenshaw · Pull Request #1690 · AllenNeuralDynamics/aind-data-schema

dougollerenshaw · 2026-01-15T21:20:18Z

Work in progress. Lots of ancillary files that we won't actually want in the repo. Remove before merging.

saskiad · 2026-01-15T21:28:31Z

examples/2026.01.16_barseq_acquisition/barseq_780346_acquisition.json

+      "dx.doi.org/10.17504/protocols.io.n2bvj82q5gk5/v1",
+      "dx.doi.org/10.17504/protocols.io.81wgbp4j3vpk/v2"
+   ],
+   "ethics_review_id": [


there is no iacuc for in vitro experiments.

Removed in c0e949b

saskiad · 2026-01-15T21:30:09Z

examples/2026.01.16_barseq_acquisition/barseq_780346_acquisition.json

+      "PLACEHOLDER_EXPERIMENTER_1",
+      "PLACEHOLDER_EXPERIMENTER_2"
+   ],
+   "protocol_id": [


why two protocols for the same thing? And why not our own protocol: https://www.protocols.io/view/barseq-2-5-kqdg3ke9qv25/v1

Those two protocols came from the methods doc that Polina shared. I didn't know about this one. Will add instead.

See 4ab930b

maybe confirm first. This is the protocol Yoh uses, so there might be a chance it's not the same?
But, the two protocols you had at first, one was an update of the other, so we should only use the one that's being used.

Polina confirmed that this new protocol is the correct one.

saskiad · 2026-01-15T21:31:31Z

examples/2026.01.16_barseq_acquisition/barseq_780346_acquisition.json

+                  {
+                     "object_type": "Channel",
+                     "channel_name": "Gene_G",
+                     "intended_measurement": "Gene sequencing - DNA base G",


this is not an appropriate intended measurement. DNA Base G is not a gene.

See change here:
4f6f2c9

saskiad · 2026-01-15T21:31:46Z

examples/2026.01.16_barseq_acquisition/barseq_780346_acquisition.json

+               "channels": [
+                  {
+                     "object_type": "Channel",
+                     "channel_name": "Gene_G",


Same as next comment: G is not a gene.

See change here: 4f6f2c9

examples/2026.01.16_barseq_acquisition/barseq_780346_acquisition.json

saskiad · 2026-01-15T21:32:59Z

examples/2026.01.16_barseq_acquisition/barseq_780346_acquisition.json

+                  },
+                  {
+                     "object_type": "Channel",
+                     "channel_name": "Gene_DAPI",


same here. Dapi is not a gene

See change here: 4f6f2c9

saskiad · 2026-01-15T21:34:27Z

examples/2026.01.16_barseq_acquisition/barseq_780346_acquisition.json

+                  }
+               ],
+               "coordinate_system": null,
+               "images": [],


this needs to be populated. This is what is being imaged.

See 3824c9e

saskiad · 2026-01-15T21:37:53Z

examples/2026.01.16_barseq_acquisition/barseq_780346_acquisition.json

+         ],
+         "connections": []
+      },
+      {


I'm not sure there should be multiple data streams here. (and I'm not sure there shouldn't, but thinking out loud here). Each round of imaging creates a new image, so I would argue each round of imaging is a separate acquisition. We could decide each round is a separate data stream - that could work.

I don't have an opinion here. Just need some guidance on how to implement whatever is most sensible.

If each round creates a distinct raw asset, then each round is a separate acquisition. We need to know how that works.

I'm not quite sure what to do with that comment. Can you help me turn that into a concrete change to make to the acquisition generator and/or a question to pose to Polina or Xiaoyin?

Are you saying that we should actually be creating three different acquistion.json files?

saskiad · 2026-01-15T21:39:07Z

examples/2026.01.16_barseq_acquisition/barseq_780346_acquisition.json

+   ],
+   "instrument_id": "Dogwood",
+   "acquisition_type": "BarcodeSequencing",
+   "notes": "BARseq acquisition of Locus Coeruleus for noradrenergic neuron projection mapping. Specimen: 51 coronal sections (20μm) spanning CCF plates 99-112. Subject 780346 received Sindbis HZ120 virus injection 22-28h pre-harvest.  BARseq experiment performed using automation template mounted on microscope. Automated microfluidics setup for reagent delivery.",


I'm generally not excited about having notes that contain the information we're trying to encode in the metadata because it enables people to develop bad habits of parsing notes instead of using the metadata properly. Likewise below.

Notes are less verbose now. I can remove them entirely if you'd prefer.

saskiad

see comments

saskiad · 2026-01-17T00:00:03Z

examples/2026.01.16_barseq_acquisition/barseq_780346_acquisition.json

+         ],
+         "code": null,
+         "notes": "Gene barcode sequencing (7 cycles) using sequential base incorporation imaging",
+         "active_devices": [


I don't think we need to list things like objective or spinning disk or filters. The things that have configs or are creating data. So cameras, lasers,

For filters, there's a field for "emission_filters" in the channel configs. See: https://aind-data-schema.readthedocs.io/en/latest/components/configs.html#channel

And they have a "device_name" field. See https://aind-data-schema.readthedocs.io/en/latest/components/configs.html#deviceconfig

As it stands, the device_names in the emission_filters align with the names in the active_devices. Is this not necessary? Should I retain the names in "emission_filters" and drop them from active_devices, or drop them in both places?

saskiad · 2026-01-17T00:00:52Z

examples/2026.01.16_barseq_acquisition/barseq_780346_acquisition.json

+            }
+         ],
+         "code": null,
+         "notes": "Gene barcode sequencing (7 cycles) using sequential base incorporation imaging",


This note feels weird to me - I would think saying something like cycle 1 of 7 makes sense, but this isn't the entire 7 cycles, is it?

That came from this line in the methods doc that Polina shared:

I'm open to any wording here or removing entirely. Let me know what you think makes sense.

saskiad · 2026-01-17T00:03:13Z

examples/2026.01.16_barseq_acquisition/barseq_780346_acquisition.json

+         "configurations": [
+            {
+               "object_type": "Imaging config",
+               "device_name": "Ti2-E__0",


I'm not sure what device this is - the main microscope? Is this the correct device to pin to? We can use the instrument itself (correct me if I'm wrong @dbirman)

That is the device name for the Microscope in the instrument:

I think the imaging config here should point to the full instrument_id in this case.

So the field name should say "device_name": "Dogwood"?

Done. See e80306e

saskiad · 2026-01-17T00:04:04Z

examples/2026.01.16_barseq_acquisition/barseq_780346_acquisition.json

+               "channels": [
+                  {
+                     "object_type": "Channel",
+                     "channel_name": "GeneSeq_G",


This is a preference thing, but I would name the channel around the color if that's possible and leave the meaning to the intended measurement.

We don't yet have the mapping for base to wavelength. And even when we do, I'm afraid there might be some subjectivity in how color names are mapped to wavelength.

I think we have 3 options:

Leave as is (channel identifies base being imaged)

Name channels with wavelength (precise, but maybe not intuitive)

Name channels with color. If we do this, I'll need guidance on how to assign color names.

saskiad · 2026-01-17T00:04:28Z

examples/2026.01.16_barseq_acquisition/barseq_780346_acquisition.json

+                  {
+                     "object_type": "Channel",
+                     "channel_name": "GeneSeq_G",
+                     "intended_measurement": "Fluorescent signal from sequencing reaction indicating guanine incorporation",


Might be better to make this less verbose.

I think "Guanine" is sufficient, personally.

Done. See b29bf11

saskiad · 2026-01-17T00:05:48Z

examples/2026.01.16_barseq_acquisition/barseq_780346_acquisition.json

+                  },
+                  {
+                     "object_type": "Channel",
+                     "channel_name": "GeneSeq_DAPI",


I'd remove "GeneSeq" since DAPI is not part of the sequencing.

Switched to just "DAPI". See 58759e6

saskiad

More comments - great job on the imaging config. I think the big question is about whether each round is a distinct raw asset or not.

…the instrument_id)

dougollerenshaw · 2026-01-17T00:48:42Z

More comments - great job on the imaging config. I think the big question is about whether each round is a distinct raw asset or not.

How do we resolve that question? Does it come down to how data is actually stored? Do we need to reach out to Xiaoyin or someone else?

dougollerenshaw added 3 commits January 15, 2026 13:15

Moved barseq folder to aind-data-schema repo

a4e8334

Removed temp file

5af4c33

Upload json file

5eacb56

saskiad reviewed Jan 15, 2026

View reviewed changes

examples/2026.01.16_barseq_acquisition/barseq_780346_acquisition.json Show resolved Hide resolved

saskiad reviewed Jan 15, 2026

View reviewed changes

dougollerenshaw added 5 commits January 15, 2026 14:06

Replaced protocol link

4ab930b

Fixed channel and intended_measurement names

153212f

Ran new acq generation code

4f6f2c9

Removed references to ethics review id

c0e949b

Used ImageSPIM

3824c9e

dougollerenshaw requested review from dbirman and saskiad January 16, 2026 23:43

saskiad reviewed Jan 17, 2026

View reviewed changes

dougollerenshaw added 2 commits January 16, 2026 16:18

change names from 'GenSeq_DAPI' and 'Hyb_DAPI' to just 'DAPI'

58759e6

Simplified intended measurement

b29bf11

saskiad reviewed Jan 17, 2026

View reviewed changes

Changed "Imaging config" "device_name" from "Ti2-E__0" to "Dogwood" (…

e80306e

…the instrument_id)

Conversation

dougollerenshaw commented Jan 15, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

saskiad left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dougollerenshaw Jan 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

dougollerenshaw Jan 17, 2026 •

edited

Loading