search

NeuroDataDesign / seelviz-f16s17

Undergraduate repository for Albert, Luke, and Tony
Apache License 2.0
1 stars 2 forks source link

Figure out how to deal with hierarchical atlas #51

Closed mlee156 closed 7 years ago

mlee156 commented 7 years ago

Figure out how to deal with hierarchical atlas - deliverable is jupyter notebook with our work

toekneesunshine commented 7 years ago

Talked to Jovo/Greg about how to do this. Greg gave us the code for removing individual regions from the atlas. After a lot of searching, I couldn't find a better CSV version of the ARA (with just the most specific regions that we're interested in). Since manually parsing 1000 regions and cross referencing them to the atlas hierarchy takes a long time, Jovo recommended that we visualize just the ARA to see if there was a pattern to the atlas hierarchy (eg: intermediary regions are cut off by their subsidiaries).

I wrote an ipynb that takes the atlas region brain and runs through a similar process to generate our atlas region plots after ndstore came back online. If you turn off the 'tag' for unknown (region ID = 0), you get the annotations.

Code here.
Visual here.
Plotly here.
CSV counts here.

toekneesunshine commented 7 years ago

Hey @kkutten1 I've looked at your code here, but I still have some questions on how to isolate the most specific layers of the atlas.

Can I meet up with you sometime to go over it?

kkutten1 commented 7 years ago

@toekneesunshine: Just sent you an email about meeting.

toekneesunshine commented 7 years ago

Met with Kwame, who helped me generate the most specific regions of the ARA. We now dynamically do so during the atlas region graph step by calling the JSON -- I've made the edits on the in the cldev repo, and we're putting it into the pipeline.

toekneesunshine commented 7 years ago

Jon and I put this into the latest version of the pipeline. We gucci.