> -----Original Message-----
> From: Charlie 
> Sent: Wednesday, February 18, 2004 8:56 AM
> To: Warren DeLano
> Subject: Re: [PyMOL] selecting multiple atoms ie oxygen
> Warren DeLano wrote:
> > John,
> >  
> >    color red, 5paa and elem o
> > 
> > The problem with using asterices as wildcards in atom names is that 
> > some ill-conceived PDB files actually use them in atom names.
> > 
> > However, PyMOL does support the use of a terminal wildcard in some 
> > cases, such as with the delete command...
> > 
> > create obj01, none
> > create obj02, none
> > delete obj*
> > 
> > And with residue names
> > 
> > color red, as*
> > color blue, gl*
> > color pink, hi*
> > 
> > Cheers,
> > Warren
> Hi Warren,
> I've struggled with this a couple of times.
> Could it be worth finding another way round, so that a 
> consistent role for * exists.
> Might it not be better to require users to somehow escape *'s 
> in atom names and then allow * as a wildcard. Although this 
> might be less clear for newbies, the lack of * as a wilcard 
> in atom names is currently unclear. I don't know much about 
> python, but the unix system of escaping special chars with a 
> backslash would be an option, then one could select 
> foo,(bar//A/50/*1\*) to get atoms O1* and C1* from the molecule.
> You've probably been through this and have a perfectly sound 
> reason for not doing it !

The newbie issue is what troubled me.  Unix hacks know how to edit PDB files
to replace asterisks with something more benign, and they know to escape
common wildcards -- but the ordinary person does not.  

Clearly we can't make everyone happy, so perhaps consistency should be the
guide?  But on the other hand, we are talking about a huge portion of the
PDB.  Nearly every nucleic acid structure seems to suffer from this
unfortunate naming convention (~5000 PDB entries contain "C1\*" according to
grep of a recent copy of the PDB).

It is true that well-established conventions already exist for handling
asterisks, but I don't believe in following conventions blindly,
particularly when so many people would be negatively affected.

I welcome further discussion on this point.  Guidance from the community
will be crucial, since I don't have a good solution in mind yet.  Some food
for thought:

1) Are atom name wildcards really needed when a more precise way of
selecting by element symbol already exists?

2) If so, then what are the proposals?

        a. Escape non-wildcard asterisks with backslash? (regexp convention,
but would trip-up newbies, break current PyMOL scripts, and inconvenience a
whole field of research)

        b. Escape wildcard asterisks in atom names with a backslash (that
would be very backwards from the standard convention and create further

        c. Add a configurable "atom_name_wildcard" toggle?

        d. Support alternative wildcards for atom names?
         Would "." or ".*" work? 

Also note that currently PyMOL doesn't have a regexp engine, and it doesn't
support full wildcards -- just terminal astericks in a few situations.  Full
regexps matching (and thus full convention adherence) would be a nice
addition in the future, but it will need to be configured somehow as well,
probably via some global setting like "regexp_based_matching".


Reply via email to