AMBER Archive (2005)

Subject: Re: AMBER: ptraj hbond analysis

From: Thomas Cheatham (cheatham_at_chpc.utah.edu)
Date: Thu Oct 20 2005 - 21:03:36 CDT


> PTRAJ: donor mask :1-54_at_O
> Mask [:1-54_at_O] represents 54 atoms
>
> PTRAJ: acceptor WAT O H1
>
> PTRAJ: acceptor WAT O H2

...

> Estimated memory usage for this hbond call: 1986.48 MB
> donors: 54 acceptors: 1604

The estimated memory usage for this command is 2GB of RAM. Likely your
machines cannot handle this...

> Processing AMBER restart file ../col.crd
> ERROR in safe_malloc: Error in alloc of 693274464 bytes

(error allocating ~700MB of memory)

I would suggest using the solventacceptor command instead as this has
significantly less of a memory footprint since only waters close to the
donor atoms are tracked.

In the script you sent, remove the acceptor commands. Also, I would
increase the solventneighbor to be something like 6.

Finally, the grid command used

   grid grid.list 8 3.0 8 3.0 24 3.0 :WAT max 0.8

is not recommended since it is too coarse (only covering a 24A grid
around your center). I normally use

   grid grid.list 100 0.5 100 0.5 100 0.5 :WAT_at_O

...however, for this to have any meaning (i.e. the grid of atom counts),
the structures have to be fit to a common reference frame. So, do an RMS
fit to the first frame about the region of interest. If you do the whole
molecule (:1-54) motion may blur the grid occupancy.

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu