AMBER Archive (2006)

Subject: Re: AMBER: amber9: SA Bomb in sa_arc

From: Giulio Rastelli (rastelli.giulio_at_unimo.it)
Date: Thu Oct 26 2006 - 05:56:43 CDT

Dear Ray,
the reason why I set fillration to 4 is that for some small ligands
the PBSA calc fails. However, I reset fillratio to 2, compiled pbsa
again, and run PBSA
again with scale=4, and aain I have the same error. I am running amber9
on a linux network opteron quadriprocessor with 32Gb central memory.
Is PBSA in amber9 so demanding in terms of memory? I have been using
fillratio=4
with amber8 and never had problems with scale=4, or even scale=5.
I guess that the PBSA code was rewritten in amber9, but can't find
any documentation on this.
All the best,
Giulio
Ray Luo wrote:
> Guilio,
>
> You're running out of memory. Please don't use fillratio =4 for such a
> large system. Reduce fillratio to 2 to see whether your memory is enough.
>
> Best,
> Ray
>
> Giulio Rastelli wrote:
>
>> Dear developers,
>> I am using amber9 to run mm_pbsa calculation on a protein-ligand complex.
>> Using a SCALE parameter of 2 (grid space 0.5), PBSA works fine.
>> When I try to increase the SCALE parameter to 4 (grid space 0.25)
>> I have this error:
>> SA surface: setting up working radii
>> SA surface: found nonzero radii 5172
>> Number of SA srf points exposed 52227
>> SA Bomb in sa_arc(): Allocation aborted 0 0
>> 1 0 0
>>
>> I am using a fillratio equal to 4.
>> Looking at the amber9 code is_init.f, it seems to me that the maxarc
>> and/or maxarcdot have to be redimensioned to higher values.
>> Before doing that, I would prefer receiving a feedback from
>> the developers.
>> Thank you very much in advance
>> Giulio
>>
>> part of MM_PBSA input:
>> ______________________________
>> @PB
>> PROC 2
>> REFE 0
>> INDI 1.0
>> EXDI 80.0
>> SCALE 4
>> LINIT 1000
>> PRBRAD 1.4
>> #
>> RADIOPT 0
>> NPOPT 1
>> CAVITY_SURFTEN 0.0072
>> CAVITY_OFFSET 0.00
>> ______________________________
>
>
> 

-- 

Prof. Giulio Rastelli
Dipartimento di Scienze Farmaceutiche
Facolta' di Bioscienze e Biotecnologie
Universita' di Modena e Reggio Emilia
Via Campi 183
41100 Modena - ITALY
-------------------------------------
tel 0039-059-2055145
fax 0039-059-2055131
www.scifar.unimore.it/on-line/Home/Gruppidiricerca/Prof.GiulioRastelli.html
-------------------------------------

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