[gmx-users] make_ndx may contain a bug?

[Paul Barrett]

Possible bug found in use of make_ndx. Sorry if it is just a mistake on
my part...

background
----------

1) I made an index file from a protein with 298 residues.
make_ndx correctly reported the following:


  1 Protein             :  3054 atoms
  2 Protein-H           :  2398 atoms
  3 C-alpha             :   298 atoms
  etc

2) I changed one residue in the protein to a non-standard residue. (I
changed THR to TPO which I had defined elsewhere to be a phosphorylated
threonine)

make_ndx came up with 


  1 Protein             :  3045 atoms
  2 Protein-H           :  2391 atoms
  3 C-alpha             :   297 atoms

Which is correct (although not what I wanted!) because TPO is not
defined as an amino acid so its carbon cannot be a C-alpha.

3) To fix this I added residue name TPO into aminoacids.dat and
incremented the line count at the top of the file. make_ndx now
correctly identified TPO as part of the protein so gave output again
exactly like (1) i.e. with 298 C-alphas. So far so good.

BUT...

4) then I added a comment line in aminoacids.dat so that locally it
looks like...


SER  
THR  
TPO  ; this is phosphorylated threonine
TRP  
TYR  


and now make_ndx comes up with...



  1 Protein             :  2609 atoms
  2 Protein-H           :  2065 atoms
  3 C-alpha             :   261 atoms

It is omitting residues.

FURTHER DIAGNOSTICS
-------------------

1) I cannot spot a pattern in which residues it is missing out: it is
not just one kind (nor even just those from the tail of aminoacids.dat)
2) most wierdly the value of C-alpha from make_ndx varies with the
contents of the comment! e.g.

COMMENT                             |       REPORTED C-ALPHA
-------------------------------------------------------------
(none)                              |            298
; this                              |            275
; this is                           |            265
; this is phosphorylated            |            261
; this is phosphorylated threonine  |            261
; (as above + 20 more characters)   |            261

I can carry on and make my simulation work but this is a nasty trap (in
a great program) that could easily catch people unawares.
3) there are other comment lines in the .dat file that do not cause
problems

many thanks

Paul Barrett