siibra-python icon indicating copy to clipboard operation
siibra-python copied to clipboard
verified publisher icon FZJ-INM1-BDA

Solidfy descriptive attributes of siibra concepts (i.e. atlas, parcellation, space, region, feature)

Open AhmetNSimsek opened this issue 2 years ago • 10 comments

Each feature name should be informative but as brief as possible especially considering the display in the explorer.

EDT: Feature names need to be reworked for the next release of siibra-explorer

AhmetNSimsek avatar Jul 12 '23 07:07 AhmetNSimsek

I would argue, the name should be concise but descriptive enough for user of siibra-python

xgui3783 avatar Jul 12 '23 07:07 xgui3783

Naming examples for queries with the region siibra.get_region("2.9", 'hoc1 left') as of 0.4a59 - dfce6eccb9bd36659125b390a8e2ca3c256ea002:

  • CellDensityProfile (Segmented cell body density) anchored at Area hOc1 (V1, 17, CalcS) Point in BigBrain microscopic template (histology) [13.452011108398438,-65.5,7.272993564605713]
  • BigBrainIntensityProfile (Modified silver staining) anchored at Area hOc1 (V1, 17, CalcS) left Point in BigBrain microscopic template (histology) [-5.160130023956299,-58.976200103759766,2.8994600772857666]
  • LayerwiseCellDensity (Cell body density) anchored at Area hOc1 (V1, 17, CalcS) Set of points BigBrain microscopic template (histology): (13.229360580444336,-63.70000076293945,6.014106750488281), (13.452011108398438,-65.5,7.272993564605713), (19.09100914001465,-64.9000015258789,-0.36307409405708313), (14.474964141845703,-60.400001525878906,8.105033874511719), (20.199174880981445,-64.5999984741211,-0.44111010432243347), (14.780386924743652,-58.29999923706055,5.0198073387146), (12.853570938110352,-65.19999694824219,8.625871658325195), (-3.9482669830322266,-42.70000076293945,9.710223197937012), (13.53404426574707,-64.30000305175781,5.984400749206543), (13.238324165344238,-65.80000305175781,8.006086349487305)
  • LayerwiseBigBrainIntensities (Modified silver staining) anchored at Area hOc1 (V1, 17, CalcS) left
  • Selected 1 micron scans of BigBrain histological sections (v1.0), section 454 (cell body staining)
  • Cellular level 3D reconstructed volumes at 1µm resolution within the BigBrain occipital cortex (v1.0) - VOI1 (cell body staining)
  • ReceptorDensityFingerprint (Neurotransmitter receptor density) anchored at Area hOc1 (V1, 17, CalcS)
  • ReceptorDensityProfile (5-HT1A receptor density) anchored at Area hOc1 (V1, 17, CalcS)

(Ebrains query)

  • CEREBRUM-7T: Fast and Fully-volumetric Brain Segmentation of 7 Tesla MR Volume

(gene=siibra.vocabularies.GENE_ NAMES.GABARAP)

  • GeneExpressions (Gene expression) anchored at Area hOc1 (V1, 17, CalcS) left Set of points MNI 152 ICBM 2009c Nonlinear Asymmetric: (-7.0,-75.0,9.0), (-7.0,-75.0,9.0), (-6.0,-86.0,10.0), (-6.0,-86.0,10.0), (-7.0,-86.0,10.0), (-7.0,-86.0,10.0), (-6.0,-94.0,5.0), (-6.0,-94.0,5.0), (-3.0,-68.0,3.0), (-3.0,-68.0,3.0), (-3.0,-62.0,2.0), (-3.0,-62.0,2.0), (-3.0,-80.0,10.0), (-3.0,-80.0,10.0), (-7.0,-99.0,2.0), (-7.0,-99.0,2.0), (-11.0,-59.0,-3.0), (-11.0,-59.0,-3.0), (-17.0,-67.0,-2.0), (-17.0,-67.0,-2.0), (-6.0,-67.0,3.0), (-6.0,-67.0,3.0), (-7.0,-67.0,1.0), (-7.0,-67.0,1.0), (-5.0,-80.0,8.0), (-5.0,-80.0,8.0), (-7.0,-99.0,-2.0), (-7.0,-99.0,-2.0), (-5.0,-64.0,-3.0), (-5.0,-64.0,-3.0), (-5.0,-65.0,-3.0), (-5.0,-65.0,-3.0), (-9.0,-86.0,14.0), (-9.0,-86.0,14.0), (-12.0,-77.0,13.0), (-12.0,-77.0,13.0), (-12.0,-86.0,1.0), (-12.0,-86.0,1.0), (-2.0,-65.0,2.0), (-2.0,-65.0,2.0), (-2.0,-86.0,-7.0), (-2.0,-86.0,-7.0), (-2.0,-86.0,-4.0), (-2.0,-86.0,-4.0), (-6.73,-67.3,-3.72), (-6.73,-67.3,-3.72), (-7.96,-74.22,7.64), (-7.96,-74.22,7.64), (-9.0,-69.0,-1.0), (-9.0,-69.0,-1.0), (-6.93,-78.81,8.71), (-6.93,-78.81,8.71), (-11.5,-67.13,0.55), (-11.5,-67.13,0.55), (-4.88,-62.41,7.23), (-4.88,-62.41,7.23), (-11.14,-96.66,0.87), (-11.14,-96.66,0.87), (0.06,-95.38,0.79), (0.06,-95.38,0.79)

For the other voi:

  • Fiber structures of a human hippocampus based on joint DMRI, 3D-PLI, and TPFM acquisitions (HSV fibre orientation map)
  • Fiber structures of a human hippocampus based on joint DMRI, 3D-PLI, and TPFM acquisitions (T2)
  • Fiber structures of a human hippocampus based on joint DMRI, 3D-PLI, and TPFM acquisitions (blockface)

Querry for julich brain 2.9 parcellation:

  • StreamlineLengths (StreamlineLengths) anchored at minds/core/parcellationatlas/v1.0.0/94c1125b-b87e-45e4-901c-00daee7f2579-290 with cohort HCP
  • StreamlineCounts (StreamlineCounts) anchored at minds/core/parcellationatlas/v1.0.0/94c1125b-b87e-45e4-901c-00daee7f2579-290 with cohort 1000BRAINS
  • FunctionalConnectivity (RestingState) anchored at minds/core/parcellationatlas/v1.0.0/94c1125b-b87e-45e4-901c-00daee7f2579-290 with cohort HCP
  • RegionalBOLD (Regional BOLD signal) anchored at minds/core/parcellationatlas/v1.0.0/94c1125b-b87e-45e4-901c-00daee7f2579-290 with paradigm rfMRI_REST2_LR_BOLD

AhmetNSimsek avatar Jul 12 '23 07:07 AhmetNSimsek

Dev meeting 07.12.2023 update:

Several attributes describing siibra concepts are not necessarily clearly defined. The table below shows what it should be and the restrictions on the attributes.

Concept Name Shortname key id description
Atlas title of a panel or a window to label UI elements autocompletion and registry at run time to encode the state  
Parcellation title of a panel or a window to label UI elements autocompletion and registry at run time to encode the state  
Region title of a panel or a window        
Parc map title of a panel or a window   autocompletion and registry at run time    
Space title of a panel or a window to label UI elements autocompletion and registry at run time to encode the state  
Feature title of a panel or a window search lists and title on side bar   to encode the state  
           
style human readable human readable upper case wo special chars uuid and persistent human readable
length 120 chrs or so: important words show up in truncated version too fits to the chips not constrained   a paragraph

see features names in version v0.4a59 see features names in version 1.0a05

AhmetNSimsek avatar Dec 07 '23 10:12 AhmetNSimsek

Names as of v1.0a05 (see comment for v.04a59)

  • CompoundFeature of 1579 BigBrainIntensityProfile features grouped by (Modified silver staining modality) anchored at Area hOc1 (V1, 17, CalcS) left with Set of 1579 points in the Bounding box from (-63.69,-59.94,-29.09) mm to (0.91,77.90,54.03)mm in BigBrain microscopic template (histology) space

    • BigBrainIntensityProfile (Modified silver staining) anchored at Area hOc1 (V1, 17, CalcS) left with Point in BigBrain microscopic template (histology) [-9.495989799499512,72.24669647216797,-12.074600219726562]

  • CompoundFeature of 10 CellDensityProfile features grouped by (Segmented cell body density modality) anchored at Area hOc1 (V1, 17, CalcS) with Set of 10 points in the Bounding box from (-3.95,-65.80,-0.44) mm to (20.20,-42.70,9.71)mm in BigBrain microscopic template (histology) space

    • CellDensityProfile (Segmented cell body density) anchored at Area hOc1 (V1, 17, CalcS) with Point in BigBrain microscopic template (histology) [13.238324165344238,-65.80000305175781,8.006086349487305]

  • CompoundFeature of 16 ReceptorDensityProfile features grouped by (Receptor density modality) anchored at Area hOc1 (V1, 17, CalcS)

    • ReceptorDensityProfile (Receptor density) anchored at Area hOc1 (V1, 17, CalcS) for 5-HT1A

  • LayerwiseBigBrainIntensities (Modified silver staining) anchored at Area hOc1 (V1, 17, CalcS) left with Set of 1579 points in the Bounding box from (-63.69,-59.94,-29.09) mm to (0.91,77.90,54.03)mm in BigBrain microscopic template (histology) space

  • LayerwiseCellDensity (Cell body density) anchored at Area hOc1 (V1, 17, CalcS) with Set of 10 points in the Bounding box from (-3.95,-65.80,-0.44) mm to (20.20,-42.70,9.71)mm in BigBrain microscopic template (histology) space

  • #47: selected 1 micron scans of BigBrain histological sections (v1.0) (cell body staining)

  • The Enriched Connectome - Block face images of full sagittal human brain sections (blockface)

  • CHENONCEAU: dti fractional anisotropy - 200um (DTI)

  • VOI2: Cellular level 3D reconstructed volumes at 1µm resolution within the BigBrain occipital cortex (v1.0) (cell body staining)

(Ebrains query)

  • Ebrains Dataset: CEREBRUM-7T: Fast and Fully-volumetric Brain Segmentation of 7 Tesla MR Volumes (v1)

(gene=siibra.vocabularies.GENE_ NAMES.GABARA)

  • GeneExpressions (Gene expression) anchored at Set of 64 points in the Bounding box from (-17.00,-99.00,-7.00) mm to (0.06,-59.00,14.00)mm in MNI 152 ICBM 2009c Nonlinear Asymmetric space

(Connectivity for Julich Brain 3.0)

  • CompoundFeature of 78 AnatomoFunctionalConnectivity features grouped by (AnatomoFunctionalConnectivity modality, ADULT PHARMACO-RESISTANT FOCAL EPILEPSY PATIENTS cohort) anchored at Julich-Brain Cytoarchitectonic Atlas (v3.0.3)
    • AnatomoFunctionalConnectivity (AnatomoFunctionalConnectivity) anchored at Julich-Brain Cytoarchitectonic Atlas (v3.0.3) with cohort ADULT PHARMACO-RESISTANT FOCAL EPILEPSY PATIENTS - P_01: amplitude__mad
  • CompoundFeature of 200 FunctionalConnectivity features grouped by (FunctionalConnectivity modality, HCP cohort, Resting state (EmpCorrFC concatenated) paradigm) anchored at Julich-Brain Cytoarchitectonic Atlas (v3.0.3)
    • FunctionalConnectivity (FunctionalConnectivity) anchored at Julich-Brain Cytoarchitectonic Atlas (v3.0.3) with cohort HCP - 000, Resting state (EmpCorrFC concatenated) paradigm -CompoundFeature of 200 StreamlineCounts features grouped by (StreamlineCounts modality, HCP cohort) anchored at Julich-Brain Cytoarchitectonic Atlas (v3.0.3)
    • StreamlineCounts (StreamlineCounts) anchored at Julich-Brain Cytoarchitectonic Atlas (v3.0.3) with cohort HCP - 000

AhmetNSimsek avatar Jan 30 '24 09:01 AhmetNSimsek

Suggestions for features

  • further reduce bounding boxes and point sets
    • Points: only show 2 digits after comma
    • BoundingBox: [(minpoint),(maxpoint)]mm on space short name
    • PointSet: ### points in BoundingBox
  • Only use modalities in names, no class names
  • IF there is an extra calculation involved, specify it (perhaps the word assigned?)
    • if there is assignment from spacial anchor to a region, specify region and pracellation
  • Pull VOI names from configuration

Examples

  • 1579 modified silver staining profiles - Area hOc1 (V1, 17, CalcS) left, 1579 points in ([-63.69,-59.94,-29.09], [0.91,77.90,54.03])mm in BigBrain
    • Modified silver staining profile of [-9.49,72.24,-12.07] in BigBrain
  • 10 segmented cell body density profiles - Area hOc1 (V1, 17, CalcS),10 points in ([-3.95,-65.80,-0.44], [20.20,-42.70,9.71])mm on BigBrain
    • Segmented cell body density profile of [13.23,-65.80,8.00] on BigBrain - Area hOc1 (V1, 17, CalcS)
  • 16 receptor density profiles - Area hOc1 (V1, 17, CalcS)
    • 5-HT1A receptor density profile - Area hOc1 (V1, 17, CalcS)

(gene=siibra.vocabularies.GENE_ NAMES.GABARAP)

  • GABARAP gene expression in 64 points in ([-17.00,-99.00,-7.00], [0.06,-59.00,14.00])mm in ICBM 2009c nonl asym

  • Layerwise modified silver staining at 1579 points in ([-63.69,-59.94,-29.09], [0.91,77.90,54.03])mm on BigBrain assigned to Area hOc1 (V1, 17, CalcS) left of Julich Brain 2.9

  • Layerwise cell body density anchored at 10 points in ([-3.95,-65.80,-0.44], [20.20,-42.70,9.71])mm on BigBrain assigned to Area hOc1 (V1, 17, CalcS) left of Julich Brain 2.9

  • Section 47: 1-micron cell-body-staining scans of BigBrain (v1.0)

  • The Enriched Connectome - Block face images of full sagittal human brain sections

  • CHENONCEAU: DTI fractional anisotropy - 200um

(Connectivity for Julich Brain 3.0)

  • 78 anatomo functional connectivity matrices for Julich-Brain 3.0.3 - cohort: adult pharmaco-resistant focal epilepsy patients
    • P_01: amplitude__mad anatomo functional connectivity for Julich-Brain 3.0.3 - cohort: adult pharmaco-resistant focal epilepsy patients
  • 200 functional connectivity matrices for Julich-Brain 3.0.3 - cohort: HCP, paradigm: Resting state (EmpCorrFC concatenated)
    • Subject 000 functional connectivity for Julich-Brain 3.0.3 - cohort: HCP, paradigm: Resting state (EmpCorrFC concatenated)
  • 200 streamline count matrices for Julich-Brain 3.0.3 - cohort: HCP
    • Subject 000 streamline counts for Julich-Brain 3.0.3 - cohort: HCP

No idea at the moment

  • VOI2: Cellular level 3D reconstructed volumes at 1µm resolution within the BigBrain occipital cortex (v1.0) (cell body staining)

No change:

(Ebrains query)

  • Ebrains Dataset: CEREBRUM-7T: Fast and Fully-volumetric Brain Segmentation of 7 Tesla MR Volumes (v1)

AhmetNSimsek avatar Jan 30 '24 10:01 AhmetNSimsek

n.b. changing names will change the feature id.

This invalidates all URLs that has feature selected.

as discussed with @AhmetNSimsek :

  • for preconfigured features
    • we will retroactively add an id attribute to the preconfigure json. This id will be derived from the id as they appear in siibra-python v0.4a<latest>
    • during ingestion, these features will use the hardcoded id as their id
  • for compound/live-query features
    • we need to check if any sane urls has any live query feature selected (likely none)
    • as such ensuring the ids stays consistent for these features are less urgent for now

Also to consider: how to more efficiently encode the id of compound/live query feature

xgui3783 avatar Jan 30 '24 10:01 xgui3783

Are the suggestions above for the name attribute?

I think we have to suggest the complete picture of all necessary attributes, building on the table from Dec 7 where we outlines the use of the different properties. I suggest to expand the suggestions above into a table, with columns "name", "shortname", "description".

One particular issue is in my eyes, whether the location information is needed in the name at all. I think the location (assigned region, coordinate[s], bounding box) could well be a separate attribute, or be requested from the str representation of the anatomical anchors when needed. The question is wether the location is required in the name to disambiguate in any of the relevant use cases.

dickscheid avatar Feb 21 '24 10:02 dickscheid

re: if location is needed at all in name and/or shortname attribute

correct me if I am wrong, but in some cases, the location is the only differentiator, e.g. big brain cell intensity (?)

ping @AhmetNSimsek

xgui3783 avatar Feb 21 '24 11:02 xgui3783

Are the suggestions above for the name attribute?

Yes since the explorer currently uses the name property to populate the text for a feature. Ideally, we should define and discuss all. However, this is currently a blocking issue for the release of next siibra-explorer version (which includes some improvements along with compound features (so F-Tract as well)). Therefore, we wanted to start "small" and go for a release ASAP.

One particular issue is in my eyes, whether the location information is needed in the name at all. I think the location (assigned region, coordinate[s], bounding box) could well be a separate attribute, or be requested from the str representation of the anatomical anchors when needed. The question is wether the location is required in the name to disambiguate in any of the relevant use cases.

There are features that has an anchor with location only. Therefore, distinguishing them is impossible without the location information. We could remedy this by indexing the points for example, i.e., "Modified silver staining intensity - point #1" (since it will be a compound feature). (Then the user can check the exact location by location property.) BUT this is not a guaranteed for features to be introduced in the future.

AhmetNSimsek avatar Feb 21 '24 11:02 AhmetNSimsek

I think we have to suggest the complete picture of all necessary attributes, building on the table from Dec 7 where we outlines the use of the different properties. I suggest to expand the suggestions above into a table, with columns "name", "shortname", "description".

Concept Name Shortname key id description
CF BigBrainIntensityProfile 1579 modified silver staining profiles - Area hOc1 (V1, 17, CalcS) left, 1579 points in ([-63.69,-59.94,-29.09], [0.91,77.90,54.03])mm in BigBrain  
GeneExpression GABARAP gene expression in 64 points in ([-17.00,-99.00,-7.00], [0.06,-59.00,14.00])mm in ICBM 2009c nonl asym  
LayerwiseBigBrainIntensity Layerwise cell body density anchored at 10 points in ([-3.95,-65.80,-0.44], [20.20,-42.70,9.71])mm on BigBrain assigned to Area hOc1 (V1, 17, CalcS) left of Julich Brain 2.9  
Sections Section 47: 1-micron cell-body-staining scans of BigBrain (v1.0)
BlockfaceVOI The Enriched Connectome - Block face images of full sagittal human brain sections
CF AnatomoFunctionalConnectivity 78 anatomo functional connectivity matrices for Julich-Brain 3.0.3 - cohort: adult pharmaco-resistant focal epilepsy patients  
AnatomoFunctionalConnectivity P_01: amplitude__mad anatomo functional connectivity for Julich-Brain 3.0.3 - cohort: adult pharmaco-resistant focal epilepsy patients  
CF FunctionalConnectivity 200 functional connectivity matrices for Julich-Brain 3.0.3 - cohort: HCP, paradigm: Resting state (EmpCorrFC concatenated)  
FunctionalConnectivity Subject 000 functional connectivity for Julich-Brain 3.0.3 - cohort: HCP, paradigm: Resting state (EmpCorrFC concatenated)  
EbrainsDataset Ebrains Dataset: CEREBRUM-7T: Fast and Fully-volumetric Brain Segmentation of 7 Tesla MR Volumes (v1)  
ReceptorDensityFingerprint 16 receptor density profiles - Area hOc1 (V1, 17, CalcS)  
ReceptorDensityProfile 5-HT1A receptor density profile - Area hOc1 (V1, 17, CalcS)  

AhmetNSimsek avatar Feb 21 '24 11:02 AhmetNSimsek