Murphy Lab

Home
Information
 People
 Addresses
 Cytometry Development Workshop
 FCS API

Research
 Projects
 Publications
 Software
 Presentations
 Flow Cytometry

Services
 PSLID
 SLIF
 Waldo

Data
 Download

Affiliations
 Carnegie Mellon University
 Lane Center for Computational Biology
 Center for Bioimage Informatics
 Biological Sciences Department
 Biomedical Engineering Department
 Machine Learning Department
 MBIC






Murphy Lab - Software - PNAS 2010 Supervised Unmixing


The software and data used for the following paper can be downloaded below:

T. Peng, G.M.C. Bonamy, E. Glory, D. Rines, S. K. Chanda, and R. F. Murphy (2010) Determining the distribution of probes between different subcellular locations through automated unmixing of subcellular patterns. Proc. Natl. Acad. Sci. U.S.A. 107:2944-2949.

The data and code are contained in six .tgz files (tar archives created with the -z option):
gnf_unmix_code.tgz7.9 MB (Matlab code)
gnf_unmix_savedresults.tgz815 MB (all intermediate results but not raw images)
gnf_images_MitoLyso1.tgz979 MB (raw images for main U2OS dataset: tar file 1 of 2)
gnf_images_MitoLyso2.tgz1.2 GB (raw images for main U2OS dataset: tar file 2 of 2)
gnf_images_MitoLysoER.tgz1.9 GB (raw images used for outlier testing)
gnf_images_drug.tgz1.7 GB (raw images used for drug effect testing)

To use these files, start by downloading and unpacking the first file above. It will create a "PatternUnmixing" folder in the current directory. It will contain some top level Matlab scripts and also a "functions" folder within it containing the rest of the code.

  • If you would like to regenerate only the figures using the saved intermediate results, also download and unpack gnf_unmix_savedresults.tgz into the current directory (it should create an "inter_results" folder that should be at the same level as the "functions" folder) and then run MasterScriptJustFigures.m. This should take less than a minute.
  • If you would just like to regenerate the intermediate results but avoid recalculating features, also download and unpack gnf_unmix_savedresults.tgz into the current directory (it should create an "inter_results" folder that should be at the same level as the "functions" folder) and then run MasterScriptFeatAlreadyCalculated.m. Only the "features" folder from within the "inter_results" folder will be used and all other files will be overwritten with new ones. This will take about 24 h on a 1.8 GHz machine.
  • If you would like to regenerate all results from scratch, also download the four raw image tar files (gnf_images_MitoLyso1.tgz, gnf_images_MitoLyso2.tgz, gnf_images_MitoLysoER.tgz, gnf_images_drug.tgz) into the current directory, unpack them (making sure that the folders from within both MitoLyso1 and MitoLyso2 end up in the /images/MitoLyso folder), and run MasterScriptWithFeatCalculation. It will generate all of the files that are in gnf_unmix_savedresults.tgz, and save the figures used in the paper into a "figures" directory.

System requirements

The scripts were written for Matlab on Linux, version 7.0 or above, and have also been tested for Matlab on Mac OS. The statistics toolbox, image processing toolbox and optimization toolbox for Matlab are also required.




Last Updated: 22 Jun 2011




Copyright © 1996-2013 by the Murphy Lab, Carnegie Mellon University