- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]

From: Paulo Costa <pjfmcosta.gmail.com>

Date: Thu, 23 Feb 2017 14:57:40 +0000

Dear AMBER users

I'm sorry for the long post ...

I'm performing some test calculations with the pbsa solver (AmberTools16)

to get the solvation free energies of small molecules. These small

molecules are built with antechamber using GAFF atom types.

I started with a simple single point calculation using the example from the

manual

Sample single point PB calculation

&cntrl

/

inp=2

&pb

npbverb=1, istrng=150, fillratio=1.5, saopt=1, radiopt=1

/

I got the non-polar contribution modeled by 2 terms, cavity + dispersion

(inp=2) and I since GAFF atom types are employed, radii from the prmtop

file will be used according to the manual (radipot=1). This calculation

yielded

EPB = -12.3409 | ECAVITY= 16.4597 | EDISPER = -15.1553

So, let's imagine I want to optimize the radii of the elements of my

molecule. The manual states that one should change the RADII section of the

topology and read those radii via RADIOPT=0. But if I use the exact same

topology as before, the radii should be the same. Therefore, I repeated the

calculation with

Sample single point PB calculation

&cntrl

/

inp=2

&pb

npbverb=1, istrng=150, fillratio=1.5, saopt=1, radiopt=0

/

and I got

EPB = -12.3409, ECAVITY= 16.4597, EDISPER = -17.0143

the value for EDISPER is completely different. Using the verbose option in

both runs, I confirmed in the output that all radii (PB and NP) are exactly

the same for both runs.

Besides EDISPER, the outputs differ in the following

< Use Tan, Yang, and Luo optimized cavity radii definition

Date: Thu, 23 Feb 2017 14:57:40 +0000

Dear AMBER users

I'm sorry for the long post ...

I'm performing some test calculations with the pbsa solver (AmberTools16)

to get the solvation free energies of small molecules. These small

molecules are built with antechamber using GAFF atom types.

I started with a simple single point calculation using the example from the

manual

Sample single point PB calculation

&cntrl

/

inp=2

&pb

npbverb=1, istrng=150, fillratio=1.5, saopt=1, radiopt=1

/

I got the non-polar contribution modeled by 2 terms, cavity + dispersion

(inp=2) and I since GAFF atom types are employed, radii from the prmtop

file will be used according to the manual (radipot=1). This calculation

yielded

EPB = -12.3409 | ECAVITY= 16.4597 | EDISPER = -15.1553

So, let's imagine I want to optimize the radii of the elements of my

molecule. The manual states that one should change the RADII section of the

topology and read those radii via RADIOPT=0. But if I use the exact same

topology as before, the radii should be the same. Therefore, I repeated the

calculation with

Sample single point PB calculation

&cntrl

/

inp=2

&pb

npbverb=1, istrng=150, fillratio=1.5, saopt=1, radiopt=0

/

and I got

EPB = -12.3409, ECAVITY= 16.4597, EDISPER = -17.0143

the value for EDISPER is completely different. Using the verbose option in

both runs, I confirmed in the output that all radii (PB and NP) are exactly

the same for both runs.

Besides EDISPER, the outputs differ in the following

< Use Tan, Yang, and Luo optimized cavity radii definition

--- > Use cavity radii in the prmtop file Indeed, if one changes the RADII section in the topology for one or more atoms, this only affects the EPB values as these are used for for the electrostatic contribution calculation. Apparently the NP radii are always the Rmin values of the FF. If so, it makes sense that the ECAVITY term is equal regardless if we use radiopt = 0 or 1 since, in both cases, Rmin values are used. On the other hand, why are EDISPER values different? If we optimize a radii for an element (via the topology), we should run a radiopt = 0 but how to compare it with the non-optimized raddii? The default calculations are performed with radiopt = 1 but that will change te contribution from EDISPER as well ... If one wishes to optimize the NP radii, used for the calculation of the non-polar contributions, we have to "hack" the source code? Or is there any other way to do it (e.g. via topology)? Again, sorry for the long post and thanks in advance for your help! Paulo Costa _______________________________________________ AMBER mailing list AMBER.ambermd.org http://lists.ambermd.org/mailman/listinfo/amberReceived on Thu Feb 23 2017 - 07:00:03 PST

Custom Search