IssuesPage202407

Known Issues in the 20240710 BUSTER (including also Pipedream, Grade2 and Rhofit) release:

• Perl messages about "Use of uninitialized value in numeric eq ..."
• Problem with the "qm" and "force_field" options
• Problem with standalone lmr (limited molecular replacement) tool
• Pymol error within buster-report with newer Python versions
• BUSTER job stalling at "crossec" call on MacOS
• Incorrect number of fitted ligands in Rhofit

• See also the Grade2 known issues page

---

Perl messages about "Use of uninitialized value in numeric eq ..."

There is a buglet in the core Perl interpreter distributed with many OS versions, that produces warning messages of the type

Use of uninitialized value in numeric eq (==) at /usr/lib/x86_64-linux-gnu/perl-base/IO/Select.pm line 68.

Although without effect on the functionality of a "pipedream", "buster-report", "rhofit" or "grade" job, they can still be disconcerting and plain ugly. The easiest solution is to surpress all warning messages from our perl scripts by adding "-X" to the perl invocation. This can be done e.g. by adding

  export KeepFromEnv_BDG_perl="yes"
  export BDG_perl="/usr/bin/perl -X"

to $BDG_home/setup_local.sh and/or

  setenv KeepFromEnv_BDG_perl "yes"
  setenv BDG_perl_arg "-X"

to $BDG_home/setup_local.csh. Another option is to point to a local (distributed) version of Perl by adding

  export KeepFromEnv_BDG_perl="yes"
  case `uname` in
    Linux) export BDG_perl="$BDG_home/.mc/linux64/3.9/bin/perl";;
    Darwin) export BDG_perl="$BDG_home/.mc/darwin/3.9/bin/perl";;
  esac

to $BDG_home/setup_local.sh and/or

  setenv KeepFromEnv_BDG_perl "yes"
  if ( `uname` == "Linux" ) then
    setenv BDG_perl "$BDG_home/.mc/linux64/3.9/bin/perl"
  else
    setenv BDG_perl "$BDG_home/.mc/darwin/3.9/bin/perl"
  endif

to $BDG_home/setup_local.csh.

Alternatively, one could run the above commands also via e.g.

  KeepFromEnv_BDG_perl="yes" BDG_perl_arg="-X" pipedream ...      # bash/sh/zsh
- or -
  env KeepFromEnv_BDG_perl="yes" BDG_perl_arg="-X" pipedream ...  # tcsh/csh

To see if your particular perl version (as used by our programs) is affected by this issue, you can use the attached test_perl.sh script. Running

  ./test_perl.sh

could return something like

##
## Testing perl environment/setup for BUSTER (Pipedream, buster-report etc)
## (for helpc contact us at buster-develop@GlobalPhasing.com)
##


  BDG_home = /home/software/GPhL/BUSTER/20240710
  BDG_perl = /usr/bin/perl (v5.34.0 - see "/usr/bin/perl -V" for more details)


### see: https://github.com/Perl/perl5/issues/18238

 interaction with IPC::Cmd::run (GH 18238) =  >>> WARNING <<< 

 $IO::Select::VERSION = 1.46
 $IPC::Cmd::VERSION   = 1.04
 Perl version         = 5.034000
                      = 5.34.0

(showing vulnerability of Ubuntu 22.04 LTS) or

##
## Testing perl environment/setup for BUSTER (Pipedream, buster-report etc)
## (for helpc contact us at buster-develop@GlobalPhasing.com)
##


  BDG_home = /home/software/GPhL/BUSTER/20240710
  BDG_perl = /bin/perl (v5.36.0 - see "/bin/perl -V" for more details)


### see: https://github.com/Perl/perl5/issues/18238

 interaction with IPC::Cmd::run (GH 18238) = OK

 $IO::Select::VERSION = 1.49
 $IPC::Cmd::VERSION   = 1.04
 Perl version         = 5.036000
                      = 5.36.0

(showing Debian 12 bookworm being fixed).

The bug was introduced (it seems) at v1.42 of Perl's IO::Select and fixed at v1.48 with this bug report.

---

Problem with the qm and forcefield options

For both BUSTER and gelly_refine the two options: "qm" and "force_field" are not working as expected. This will be fixed in the next release.

Until then a work around exists if there is only a single uncharged ligand. Simply append an underscore to the three-letter code i.e.

gelly_refine -qm I0G_ ...

and it should work.

---

Problem with standalone lmr (limited molecular replacement) tool

When using a stand-alone BUSTER distribution, the lmr command throws an error

Use of uninitialized value $mtzana in concatenation (.) or string at ...

This does not happen for a combined installation of all our software (as available to Global Phasing comnsortium members). The solution is to change line 233 in $BDG_home/scripts/lmr from

$mtzana = IPCcmdcanrun( $ENV{"BDG_home"} . "/autoPROC/bin/$archname/mtzana" );

to

$mtzana = IPCcmdcanrun( $ENV{"BDG_home"} . "/autoBUSTER/bin/$archname/mtzana" );

---

Pymol error within buster-report with newer Python versions

If you get an error when running buster-report

ModuleNotFoundError: No module named 'imp'

you probably need to install e.g. on Ubuntu/Debian systems:

  sudo apt install python3-zombie-imp
 

---

BUSTER job stalling at "crossec" call on MacOS

When refining a model with hydrogens using CCP4 on MacOS (and probably only with newer Apple Silicon chips), "refine" might stall at teh call to the CCP4 program "crossec" due to some numerical issue. Until this is fixed in CCP4, you can skip the use of crossec by adding

   refine ... AutomaticFormfactorCorrection=no AutomaticFormfactorCorrectionAnalyse=no ...

to your command-line.

---

Incorrect number of fitted ligands in Rhofit

When running Rhofit using

  rhofit ... -allclusters ...

the number of fitted ligands placed into the final merged.pdb file can be incorrect in cases where several binding sites are encountered. The fix is to change line 2035 in $BDG_home/scripts/rhofit from

    $clusters_to_search = $#clusters;

to

    $clusters_to_search = $#clusters + 1;