Linux installation


I have used CellProfiler on both windows and MacOS successfully before and switched to linux (Fedora 12; 64-bit) just recently. I managed to install CellProfiler using the pre-compiled Linux version. However, I don’t know how to install CellProfiler Analyst 2.0 from the source code. I have python 2.6.2 as well as numpy 1.3.0 and scipy 0.7.0 installed. I tried ./configure in the CPAnalyst directory but that didn’t work.

Perhaps somebody could post a brief manual for the linux installation procedure.



Hi David,
Thanks for your interest. We really need to get some information out there for Linux users and developers. I just added a file DEVELOPERS.txt to the repository which has the following information.

The CellProfiler Analyst project is developed in Python and tested under Mac
OSX 10.5, Windows XP, and to a lesser extent some versions of Linux.

Code may be checked out from our SVN repository here: … CPAnalyst/

To run the developer version, you will need Python 2.5 and the following python

  • wx 2.8.10
  • MySQLdb 1.2.2
  • pysqlite2 (if using SQLite)
  • numpy 1.3
  • scipy 0.7
  • matplotlib 0.98 (for plotting functionality)
  • PIL

OPTIONAL: Classifier can be sped up by installing the user-defined function
found in mysql_plugins/classify.c. To add this function to MySQL, compile
classify.c to, and put this in a directory on the LD_LIBRARY_PATH
for MySQL, then execute this statement within MySQL:

CellProfiler Analyst can be run from a terminal as follows:

…I hope this answers your questions. I’m not 100% sure, but you might need to install python 2.5… although it might work to some extent under 2.6, so you can always try that first. I may have forgotten an intermediate step or two, so let me know if you get stuck.

Good luck!

I believe it is possible to run CPA in Linux 2.6, though a few changes may be necessary. The python version of CellProfiler runs in 2.6 without any 2.6-specific changes.

Hi, I’m following up on this old thread…

Since the installation on linux is fairly complicated. Are you planning to release binary versions of cell profiler and cell profiler analyst for linux any time soon? And if so, for which linux distributions?



Hi David,

The short answer to the CP side is “yes.” That’s something that we’re working on currently. The main focus will be on CentOS (5, then 6), with a focus on 64-bit systems (the assumption being that the majority of linux CP-users are looking to do batch processing). After that, we may try to build a .deb, likely for a newer Ubuntu distribution. Since (based on old posts) you seem to be using Fedora, this may not fit your needs perfectly–Fedora is a little more difficult to maintain packages for due to its bleeding-edge nature. While its possible a CentOS rpm would work on some variety of Fedora, I couldn’t guarantee it.

Barring all other options, you can resort to using a chroot (likely of CentOS) or various virtualization methods (virtualbox, for example). There are various issues associated with this, though, so it’s not perfect.

The shorter answer for CPA is “no.” There isn’t really any support for that at the moment. You can always try dumping the source and seeing if it runs, but it is uncharted territory at the moment.

Which OS are you working on/do you want to be working on (and what is your main application–are you just running on the desktop, or on a cluster)?



thanks for the information. I’m still using fedora but I might give CentOS or Ubuntu a try. I would use CP on this computer just as a Desktop application for pipeline development, not really for analyzing larger datasets. I have not yet used CPA but I would like to get started with it.