kuen wu

[Mac] run Phenix+Coot with Apple-Silicon Chip

Since Apple announced the silicon chip to replace the Intel CPU, issues come up with the supports of Linux-based software for biomolecular structural biology. For example, NMRPipe and Sparky-NMRFAM for NMR users are unable to run well (at least I have problems to run them on my Macbook, running Ventura 13.6).

Software for biomolecular crystallography is less a concern though it still took me a while to get them running on my Macbook with a M2 chip. I figured out a sequence to install PyMOL, Phenix, XQuartz, and CCP4 (and coot included).

  • First and optional, have a PyMOL installed. (I have my own license).
  • Download and install Phenix
  • Download and install XQuartz.
    It is noting that I installed X-Quartz 2.8.5 though CCP4 website suggests 2.8.0. The XQuartz 2.8.0 is too old as it was released in 2020. I have tried 2.8.0, 2.8.3, and 2.8.5, then I found 2.8.5 works with Phenix coupling with Coot quite well.
  • Download CCP4, unpack the compressed file. Move CCP4 to a good location (not Download). Then run “BINARY.setup” in the terminal to set up CCP4 automatically.
  • The default Mac terminal is zsh and I switched to tcsh as my old scripts are all tcsh-based. (I don’t think this is a factor, but noted here).

After the installation is completed, I test Phenix with MR, Autobuild, Phenix.refine, and phenix.reflection_file_converter (convert sca to mtz, command-line only). The autobuild app with 5 assigned processors worked superb. The Macbook is very quiet and it finished autobuild job for a 1.6Å crystal map in 35 minutes.

Here are some snapshots of Phenix including main window, autobuild, refinement, and most important “Coot”.

Coot can be directly called from the “Open in Coot” button in the refinement job. The Molprobity analyzed information including clashes, missing atoms, and violations. It is also doable to real-time point to the violated atoms in the coot from Phenix refined window.

Now it’s time to use the powerful macbook with M2 Max chip for daily data analysis.

(Note: I also test cryoEM map and the model refinement, works perfectly.)