On 15/07/11 16:05, Ed Pozharski wrote:
First of all, let me say that having coot packaged for Ubuntu is fantastic. Not only it makes it easier to install, to me another important consequence is the the Linux small-size promise is thus fulfilled.
I believe that Morten is using 0.7-pre for the packaging. However, he turns off guile scripting (it uses old dependencies) and (only) python is used instead. I don't run Coot like this, so I am somewhat unfamiliar with the details - it's supposed to work though.
However, I ran into some trouble with ligand incorporation. Specifically, malonate ion. Scenario 1. CCP4 is not configured. Nothing works. If I try to search for either 3-letter code or by keyword I am told that CCP4_BIN is not set etc., it offers to search entire disk for libcheck, which takes so long I canceled it. Attempted workaround: I put symbolic link to libcheck into /usr/local/bin or define libcheck alias. Still doesn't work - the relevant line in coot terminal may be # ERROR: can"t open (lib) list/mon_lib_list Setting CCP4_LIB to point at /usr/share/refmac doesn't help.
OK, a quite recent change is that (at the request of Keith Wilson) Coot does not use libcheck to generate 3D coordinates - the coordinates are read straight from the cif dictionary file. It might be that Morten has not yet included this - or that the python version has yet to catch up with this mechanism (it should).
Scenario 2. CCP4 is configured (6.1.13). You can find malonate through coot menu, but libcheck fails with a different message # WARNING : monomer:MLI - not found in the library . This seems to suggest that libcheck is looking in CCP4_LIB and the monomer library that comes with 6.1.13 has no malonate. If I choose methyl-malonic acid (DXX) it works fine because DXX is in 6.1.13 monomer library. So it seems that the problem is that "Ubuntu Coot" looks for monomers in /usr/share/refmac but then invokes libcheck which looks elsewhere.
OK, it shouldn't do that.
BTW, copying MLI.cif to $CCP4_LIB/data/monomers/m doesn't help, perhaps because libcheck relies in mon_lib.list or such. I can probably figure out which files to amend to include MLI, or maybe an update is available for the latest refmac monomer libs, but this would still be an ugly workaround.
I use the 5.29 dictionary File -> Search Monomer Library -> "Malonate" -> Search Then click on "MLI: Malonate Ion" -> It appears immediately That is what is supposed to happen. That's how it works for me.
I think ideal solution would be to have refmac monomer libs adopted as a separate package by CCP4, but I doubt that CCP4 is interested in maintaining packages for various linux distros. I can understand why CCP4 doesn't go the "Linux way" with dependencies - their goal is to assure successful installation, not to maximize library sharing.
Yes.
Alternatively, Ubuntu Coot could depend on a separate libcheck package - but unless I missed something, it doesn't do that now. Is this thus a bug and what could be a workaround? The only one I know is to import cif-file and ligand pdb manually, which works but makes get-monomer less useful.
If libcheck is in the path when coot is started, it should just work. libcheck doesn't need to be in a package for that to be the case.
Hmm... I'm a bit in the dark - I think that I should build and run coot without guile, like Morten does.
Paul.
