October 5th 2016 Deliverables

[mlee156]

Link to presentation: https://neurodatadesign.github.io/seelviz/reveal/week_105.html

Big Picture: By the end of this week give Ailey a pipeline that is useful and easy to use/modify Memorize that 1 voxel is 2 bytes

Note: Edited by Albert below the hash-line to specify delegation of tasks for "Show the 40,000 brightest points in registered space relative to the registered region"

Tasks:

Jon:

• [x] Complete pipeline and make Python package. Pipeline code should be pip installable.
• [x] Provide evidence that the pipeline works. Testing here.
• [ ] Complete documentation and organize Github repository

Daniel:

• [ ] Recheck Scree plots and eigenvalues to make sure they are accurate
• [ ] Extend x axis for the degree chart
• [ ] Clustering coefficient should not go above one - Make sure to re-plot so that the clustering coefficient does not exceed one

Luke:

• [x] Create Docker for pipeline with Albert once pipeline is complete https://hub.docker.com/r/lkzhu1/ubuntu/
• [x] Show the 40,000 brightest points in registered space relative to the registered region with Albert, Tony

Tony:

• [x] Address Davi Bock's issue - make a plotly plot of Cocaine 174 where voxels are colored according to component size
• Jupyter Notebook containing a generalized Plotly plotting script for density.
  • Deliverable was to apply it to Cocaine174 using ball size of 25, to color the voxels according to component size. Here is a 5000 point downsampled version I made initially (verified accuracy by finding max brightness values and comparing them to the node values on the chart). Here is the 40k node false color version I made. I also checked the maximum brightness value and compared it to the brightness I got by scanning the list.
  • For cross referencing, made a heat map coded for each 5%; shown here is the corresponding edge to color mapping for the downsampled Cocaine174. Here is the corresponding edge to color mapping for the full scale Cocaine174 plot.
• [x] Show the 40,000 brightest points in registered space relative to the registered region with Albert, Luke

Albert:

• [x] Create Docker for pipeline with Luke once pipeline is complete https://hub.docker.com/r/lkzhu1/ubuntu/
• [x] Show the 40,000 brightest points in registered space relative to the registered region with Luke, Tony Deliverable evidence listed below
• [x] Set up package page for prototype with installation instructions https://github.com/alee156/clviz Build passes on travis-ci https://travis-ci.org/alee156/clviz

"Show the 40,000 brightest points in registered space relative to the registered region with Luke, Albert, Tony" is a task for Luke, Albert, and Tony: The following is the splitting of tasks

Albert:

• [x] Downsample or obtain atlas and one raw image so that they are less than 2 GB in size Learned from Kwame how to downsample effectively using "spacing". https://github.com/NeuroDataDesign/seelviz/blob/gh-pages/miniatlas.nii https://github.com/NeuroDataDesign/seelviz/blob/gh-pages/Ndreg%20Simple%20Downsampling%20Method.ipynb
• [x] Write code for aligning atlas to the raw images instead of other way around https://github.com/NeuroDataDesign/seelviz/blob/gh-pages/ReverseAlignmentDownsampled.ipynb https://github.com/NeuroDataDesign/seelviz/blob/gh-pages/albert/brainalign.py https://github.com/NeuroDataDesign/seelviz/blob/gh-pages/albert/reverse_affine_annotations.png
• [x] Write code used for resolution 5 images on server https://github.com/NeuroDataDesign/seelviz/blob/gh-pages/albert/brainalign.py

Luke:

• [x] Run code on resolution 5 images using server https://github.com/NeuroDataDesign/seelviz/tree/gh-pages/albert
• [x] Create plotly plot representing each region of the brain as a different color https://neurodatadesign.github.io/seelviz/reveal/html/graphmlhtmls/Control258localeq.region_color.html
• [x] Sum intensity in each region https://neurodatadesign.github.io/seelviz/reveal/html/Control258localeq.region_activity.html https://neurodatadesign.github.io/seelviz/reveal/html/Control258localeq.region_activity_blackbg.html

Tony:

• [?] Create plotly plot of graphml network on new brain
  • Jupyter notebook code for raw graphML to 3D plot with edges.
  • Control258 40k nodes.
  • Aut1374 100k nodes.
  • Need the new raw .nii files, this graphml was still captured from the old .nii
  • Data still not fully propogated for other channels for the new brains
• [x] Setup server and install all packages
  • Procured and setup a Google Cloud instance (similar to an Amazon AWS EC2 instance) free trial with $300 credits.
  • CloudImage
  • SSH
  • DockerImage

[mlee156]

@jovo @gkiar

[jovo]

@all - please make a detailed list of all the improvements/adjustments so we are all on the same page.

[mlee156]

@gkiar @jovo the issues have been updated!

[jovo]

i'm not quite sure what "Address Davi Bock's issue" means? how address? what issue?

" Clustering coefficient should not go above one" true. what will you do?

[mlee156]

@jovo @gkiar I updated the issues again!

[jovo]

better!

[gkiar]

@alee156 @toekneesunshine @lukezhu1 be more specific please - how many of the brightest points? What does "useable form" mean for downsampling the image? @jonl1096 does making the python package mean you will have it on pip? Such that I can clone and then run python setup.py install? What is your DoD here?

[mlee156]

@gkiar I updated the deliverables again!

[gkiar]

👌

[jonl1096]