back to homepage
MitoSpindleScreen code

After working with Gohta Goshima and Nico Stuurman from the Vale lab on the spindle length project,  we decided to continue the collaboration and extend our work to a genome wide screen. You can read about this screen in a recent Science article here.

Below you can find all the source code that was used for image storing and analysis in the MitoSpindleScreen project.
For the web interface that was used in the screen, please check out Nico's excellent tool: phplabware

This is an overview of the analysis pipeline:



This pipeline was implemented using multiple matlab scripts and functions that perform the different analysis steps.
Overall the code includes many aspects of the computational infrastructure needed for a genome wide screen:

• Database utilities
• Segmentation
• Classification
• Gallery generation 
• Graphical User Interface for phenotype assignment
• Quality assurance (QA) and html reports
• Quantitative measurement and statistics
  
• Data explortation

html docs of all the matlab source code is here.

An overview of how database table structure can be seen here.

The code was developed for the purpose of this screen only, which has pros and cons. We made many 'shortcuts' because we were concentrated on a specific solution and did not aim to create a generatl tool for genome wide screens. Because of that, our >6500 lines of code will to be hard to reuse. To help with its future reusability I tried to organize the files into meaningful folders, rename some files, and add documentation where appropriate.  That might have 'broken' a few things, but since I don't expect this code to be usable 'as is' in any case I thought that would be better then just a pile of unreadable code...

a tarball of all m-files is here.

For the biological results of this screen, go to: http://rnai.ucsf.edu/mitospindlescreen