AMBER Archive (2009)

Subject: Re: [AMBER] number-of-EPS-exceeds-MAXESP

From: FyD (fyd_at_q4md-forcefieldtools.org)
Date: Sun Jan 11 2009 - 01:00:40 CST

Dear Karol,

> 1) "I would strongly suggest you comparing charge values using IOp(6/33=2)
> and (6/33=2,6/41=10,6/42=17)..."
>
> I have already done it - but only for one conformation.

The difficulty here is the choice of the conformation to be selected:
I mean _if_ you are interested in thinking about this problem...

> In case of - IOp(6/33=2) = Fe charge = +0.08.

Is it a iron-II or iron-III ?

> 2) "comparing charge values for different molecular orientations.."

Just to look at the "stability" of the charge values, duplicate the
Mol_red1.p2n file in R.E.D.:
  Mol_red1.p2n a single orientation "1"
  Mol_red2.p2n a single orientation "2
  Mol_red3.p2n a single orientation "3"
  Mol_red4.p2n a single orientation "4"
& run R.E.D.

> Well I have been thinking about it but based on the literature I found
> that at least for one
> "type of heme" (to which HIS and MET are linked) there are "strict"
> geometric
> dependencies= bond lengths,angles around Fe which denote optimized
> geometry.
> That's way I am not sure if charges (especially for Fe and atoms close
> to Fe) should be
> calculated based on many conformations.

Difficult to say.

> Besides I think that in case of
> heme=IRON the
> "IOp parameter" is a crucial thing - however this is only my opinion
> and I could be wrong...
>
> 3) Now about this RESP recompilation.
> Well I don't know FORTRAN. I checked in the web how I should define the
> integers and so
> on.. so I tried:
> - 2i5 ---> 2i6 - *segmentation fault*
> - 2i5 ---> 2i7 - *segmentation fault*
> so I asked one person who advised:
> - 2i5 ---> i5i6 - finally RESP worked and charges were fine.
>
> But I would like to be sure: does this change "2i5 ---> i5i6" is in
> agreement with whole
> RESP algorithm? I mean: I changed only "ESP points limit", besides RESP
> works as "default 2i5"?

I think this is just a question of format here.
I do not know fortran as well ;-)

> 4) I can also send you the last version of R.E.D.-III.1 if you wish).
>
> Well - RED was my first option but I had to use cc-pvtz with GEN keyword
> and I didn't
> know how to put cc-pvtz function in RED code + I wasn't sure if I can
> modify your code.
> But still I would be grateful if you can send me R.E.D.-III.1.

In fact, we are going to release it tonight/tomorrow as we finished to
incorporate a last feature. As you are a R.E.D.-III user, you will get
an advertisement email.

Sure we use the Gen keyword in R.E.D.: Simply look at your Gaussian
input (what you want to reach), look at the R.E.D. source code & add
what you need in the section "SECTION FOR MEP COMPUTATION" (Gaussian
input section) . Do not stay block by what is proposed in R.E.D.
create your own R.E.D. with the tricks you found in your computations...

Do you need to read the .chk file to save you computer time for
instance ? Just code it in R.E.D. (you have the variables of molecule
$NM, conformation $NC and orientation $w: JOB2-gau_m$NM-$NC-$w.com")

For "Gen", you will need to jump a line:
                        print JOB2_FILE "\n";

Then, you add the basis set description in the input
                        print JOB2_FILE "blabla 1 for iron\n";
                        print JOB2_FILE " [etc...] \n";
                        print JOB2_FILE "blabla n for iron\n";

the trick is to not mix up the iron radius set to 1.8 in R.E.D. with
the "Gen" basis set description after the Cartesian coordinates...

We are going to propose new options in R.E.D. Server/R.E.D.-IV using
such tricks (however not in the first release I think).

regards, Francois

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber