Dijk van Marc wrote:

A PDB file is essentially a table. One line from such a table can be as follows:

ATOM 14 CA GLN A 2 -27.648 -9.581 30.325 1.00 10.00

In the language description I would like to use a regular expression construct that matched 'GLN' only if the line starts with either 'ATOM' or 'HETATM'. This seams like a conditional regular expression but my attempts to implement it at such have failed so far.

Can anyone help?

Is this what you mean?

^(ATOM|HETATM).*GLN

This will match any line which starts with either 'ATOM' or 'HETATM', followed by any string of arbitrary characters, followed by the string 'GLN'.

On May 22, 2009, at 12:56 PM, Dijk van Marc wrote:

The situation: A PDB file is essentially a table. One line from such a table can be as follows:

ATOM 14 CA GLN A 2 -27.648 -9.581 30.325 1.00 10.00

In the language description I would like to use a regular expression construct that matched 'GLN' only if the line starts with either 'ATOM' or 'HETATM'. This seams like a conditional regular expression but my attempts to implement it at such have failed so far.

Start a begin/end match with a look-ahead on ATOM|HETATM:

(?:ATOM|HETATM)

And an end of $ (end of line). Inside the pattern rules have a match for the GLN.

You could use a conditional regex to do this if it was before the ATOM, but you can't do look-behinds of an unspecified length.