VMD has a rather powerful atom selection language available. It is based around the assumption that every atom has a set of associated with it values which can be accessed through keywords. These values could be boolean (is this a protein atom?), numeric (as in the atom index or atomic mass), or string (the atom name). The values can even be referenced via a Tcl array.
To start off, here are some examples of valid selection commands in VMD. Following these will be a more in depth description of how selections work.
name CA resid 35 name CA and resname ALA backbone not protein protein (backbone or name H) name 'A 1' name 'A *' name "C.*" mass < 5 numbonds = 2 abs(charge) > 1 x < 6 and x > 3 sqr(x-5)+sqr(y+4)+sqr(z) > sqr(5) within 5 of name FE exwithin 3 of protein protein within 5 of nucleic same resname as (protein within 5 of nucleic) protein sequence "C..C" name eq $atomname
There are two types of selection modes. The first is a keyword followed by a list of either values or a range of values. For example,
name CAselects all atoms with the name CA (which could be a C or a calcium);
resname ALA PHE ASPselects all atoms in either alanine, phenylalanine, or asparagine;
index 5selects the 6th atom (in the internal VMD numbering scheme).
VMD can also do range selections, similar to X-PLOR's `:' notation:
mass 5 to 11.5selects atoms with mass between 5 and 11.5 inclusive,
resname ALA to CYS TYRselects atoms in alanine, arginine, asparagine, aspartic acid, cystine, and also tyrosine.
The keyword selection works by checking each term on the list
following the keyword. The term is either a single word (eg,
name CA) or a range (eg
resid 35 to 90).
The method for determining the range checking is determined from the
keyword data type; numeric comparisons are different than string
comparisons. The comparison should work as expected so that ``8'' is
between ``1'' and ``11'' in a numeric context but not in a string one.
This may lead to some peculiar problems. Some keywords, such as segname, can take on string values but can also be used by some
people as a number field. Suppose someone labeled the segname
field with the numbers 1 through 12 on the assumption that they are
numbers. That person would be rather confused to find that
segname 1 to 11 only returns two segments. Also, strings
will be converted (via atof()) to a number so if the string
isn't a number, it will be given the value of 0. It is possible to
force a search to be done in either a string or numeric context
Selections can be combined with the boolean operators and and or, collected inside of parenthesis, and modified by not, as in
(name CA or name CB) and mass 12 to 17which selects all atoms name CA or CB and have masses between 12 and 17 amu (this could be used to distinguish a C-alpha from a calcium). VMD has operator precedence similar to C so leaving the parentheis out of the previous expression, as in:
name CA or name CB and mass 12 to 17actually selects all atoms named CA or those that are named CB and have the appropriate mass.