Dataset not showing in siibra explorer when clicking brain region

[PolarBean]

I noticed that this dataset card says that it is in the "dorsal lateral geniculate nucleus (Waxholm Space Rat Brain Atlas)" https://search.kg.ebrains.eu/instances/Dataset/d95c740d-80fe-46ad-bd75-21f53ca3b892

But when I go into the siibra waxholm atlas and visit that region I cant find it in the linked datasets. Funnily enough it shows up if I visit the thalamus in the Waxholm version 3 atlas. Also the dataset says that its linked to "primary somatosensory cortex (Waxholm Space Rat Brain Atlas)" which is not a brain region in any Waxholm atlas. However Primary Somatosensory Area is a hierarchical region in version 4.

Thank you in advance for your help with this!

[xgui3783]

Also the dataset says that its linked to "primary somatosensory cortex (Waxholm Space Rat Brain Atlas)" which is not a brain region in any Waxholm atlas.

This we will have to rely on KG curators from Oslo. Perhaps @UlrikeS91 ?

Funnily enough it shows up if I visit the thalamus in the Waxholm version 3 atlas.

(actually also shows up in waxholm v2 thalamus)

This is an issue in two parts:

1/ we are still relying on data from kg v1 to parse anchored region (we have a snapshot of kg v1, so we do not query kg directly). From the last snapshot we took of kg v1, the said dataset is anchored to Thalamus (v2) and Neocortex (v2), which leads to ...

2/ we are currently using fuzzy matching of region name + species to regions. Which is why rat thalamus matches to both v2 and v3 thalamus.

I have started a discussion at https://github.com/FZJ-INM1-BDA/siibra-python/discussions/336 to discuss moving to a more strict region matching.

[PolarBean]

Thanks for the reply. Ulrike is away on maternity leave for now so I will raise this with the curation team. One thing I'm curious about is if a dataset is registered to the v2 thalamus and thalamus doesn't exist in v4, how can this be dealt with. Is there an correspondance feature in KG? as in does v4 region x correspnd to v2 region z and therefore datasets registered to v2 can be shown in v4?

[xgui3783]

I will have to take a look in how kg v3 registers parcellation entity version. it may already be doing that, but because siibra-python is not doing any parcellation entity version matching in kg v3 at the moment, it might be missing.

[Majpuc]

Thank you for your input @PolarBean. Ulrike and I started working on this issue for some time ago and have already had several meetings with Xiao. We have a plan on how datasets should be visible in all the atlas versions. As far as understand this will be taken care off in future releases of siibra.

[xgui3783]

@Majpuc I think one of the issue @PolarBean was referring to was that, on KG, https://search.kg.ebrains.eu/instances/Dataset/d95c740d-80fe-46ad-bd75-21f53ca3b892 , one of the brain region referenced was primary somatosensory cortex (Waxholm Space Rat Brain Atlas) , which is not a region either of us are aware. This is not an issue which the siibra team can address.

Otherwise, your assessment is correct. Personally, I would am pushing for integration of siibra with KG, (in this instance, region instance with PEV).

[lzehl]

@Majpuc and @PolarBean the "primary somatosensory cortex" is a parent structure within the WHSSD Atlas that was introduced for grouping all primary somatosensory areas (which do exist in the WHSSD label file). It is true though that the release of all WHSSD Atlas data misses (as far as I know) to provide any parent structure and hierarchical relation between areas within the label file. The parent structures registered in openMINDS were though discussed in Oslo (Ulrike + @aeidi89 + Trygve, I assume) and based on this discussion introduced in the terminology registration in openMINDS of the atlas. We can of course delete them again and only register the WHSSD areas as a flat list, which I think is not beneficial for the atlas. Please let us (openMINDS) know what the preference is here. Note that from the usability perspective parent structures are quite need if more than one region of the same parent structure is studied, or if location information can only be provided on a coarser level. Possible consequence of getting rid of parent structures (in EBRAINS): recuration of all affected datasets; this might even lead to the point that datasets are not going to be tagged with WHSSD Atlas areas anymore (because only the parent structures would have worked as correct reference).

@xgui3783 note that the "primary somatorsensory area" is only registered as PE, not as PEV. Note also that the parents are not yet registered in the child structures (under "has parent" for the WHSSD Atlas, because of some missing feedback for some of the relations. @aeidi89 we should tackle this as soon as possible (if the decision is made that parent structures should be kept for the WHSSD Atlas). Ulrike was trying to solve these issues but as far as I know was still waiting on confirmation for some relations (especially across the versions of the WHSSD Atlas).

I hope this helps solving this issue. Please get in touch with me and @aeidi89 for further questions.