movement_validation bulletin

[Michael Currie]

A few of us thought it would be useful to give an occasional (perhaps weekly) email update to aid coordination and motivation in the movement_validation group.

Background

As most of you know, movement_validation is a sub-project of OpenWorm designed to verify that the simulated worm is moving "correctly" by comparing the simulation to videos of real worms.

By conducting most correspondence through this group, we can be sure to maintain a searchable historical record that may aid us or others working on the project in the future.

The repository we are working on is here: https://github.com/openworm/movement_validation

Bi-Weekly General Meeting

Last Wednesday 10 September I gave an update on behalf of our group to the OpenWorm team at our bi-weekly meeting.  You can see the note here or listen to the update I gave here (I cued it up to our part).

From DropBox to Google Drive

I've just now created a Google Drive folder for the team.  It is a sub-folder of the overall OpenWorm Public Google Drive folder.  Feel free to add or edit anything.

Separately, Jim has recently migrated our example data from DropBox to Google Drive:

• The folder is actually a sub-folder of the above team folder.
• I'm modifying the installation instructions today to reflect the changes.
  
• I'll be removing and un-sharing the shared DropBox folder, worm_data, soon.

Sincerely,

Michael

[Stephen Larson]

Thanks for this, Michael!

Project coordinator

http://openworm.org

"Talk is cheap, show me the code" - L. Torvalds

[Rick Gerkin]

Thanks for the overview and docs, Michael.  

1) "Code Progress Chart" seems to refer to three separate documents:  

a) A broken link to a pdf hosted on GitHub in the the bi-weekly agenda meeting notes.  

b) A spreadsheet I can edit:https://docs.google.com/spreadsheets/d/1dW1ukYlTu4vbm35bkf8MIZ3obP37yrKFz12X84ukOT

c) A spreadsheet I can't edit:  
https://drive.google.com/folderview?id=0B9dU7zPD0s_LdHRndU9QQ3NTRUE

Which one is authoritative going forward?  I mean I assume the editable one is the master and the others are being periodically re-published, but I want to make sure.  

2) Considering the columns of that document, I'm assuming that SciUnit will appear in movement_validation somewhere after "Experiment", "Video", "Contour" and "Pre-features", and becomes relevant either in "Features", "New Features" or maybe as late as "Statistics", and then continues through "Fitness Function" and possibly "Other Components".  

3) The code used to compute features and the code used to evaluate them should be cleanly separated.  Right now I'm not sure that is the case, but if someone can let me know the intended use of e.g. the statistics module that would be great.

4) More when I get back from Disneyworld.

[Michael Currie]

Hi Rick,

Thanks for this feedback.

1. I originally created "Code Progress Chart" as an Excel spreadsheet, which for greater accessibility I also saved as a PDF, and stored it in the movement_validation repository.  I linked to this one in old meeting minutes.

I recently converted it to a Google Spreadsheet so that it could be more easily edited in a group setting.  As a consequence I moved it to this folder.

Thus this one should be the authoritative one and the only one going forward.

I've just now changed its editable settings to include the whole movement_validation team (originally it was only the OpenWorm Core team that was set to be able to edit).  Let me know if you have any further problems.

2. Yes I agree.  The raw video needs to be converted into something more useful before we can consider it to be the "observation".  The "model" is just the contour wireframe of the simulated worm.  These are on cells G9 and G10, respectively.

3. By evaluate I assume you mean compare observation features to model features?  If so, that step is the Statistics step, and is summarized by the fitness function, which will likely be some weighted average of the statistics.

4. Have fun!

Sincerely,

Michael