R.E.D. III
Frequently Asked Questions
Below, you will find a summary of FAQ gleaned from the net. Most of them come from the AMBER (and GAMESS) mailing lists.
Questions are written in green color and the corresponding answers in black...
Please check also the AMBER mailing list archives
(http://structbio.vanderbilt.edu/archives/amber-archive/ and
http://amber.ch.ic.ac.uk/archive/)
and AMBER Web site (http://amber.scripps.edu/Questions/resp.html and
http://amber.scripps.edu/Questions/amber-archive.html)
for other information about Ante_R.E.D., R.E.D. and X R.E.D.
1) Announcement of the release of the R.E.D.-III tools
2) R.E.D.-III problem
3) running_RED-III
4) running_RED-III again
5) running_RED-III again & always
6) Reorientation information
1) Announcement of the release of the R.E.D.-III tools
From: FyD <fyd@q4md-forcefieldtools.org>
Date: April, 13th, 2007
Please, click here
2) R.E.D.-III problem
From: FyD <fyd@q4md-forcefieldtools.org>
Date: 8 Jul 2007
Quoting Jarrod Smith
I'm trying to use R.E.D.-III to calculate partial charges of some modified DNA nucleotides. For starters, I've been seeing how closely I can reproduce the AMBER charges for the four standard bases to make sure my protocol is working OK.
The slightly modified DA.com and Mol_red1.p2n files were created from your Ante-RED script. I've run the Gaussian03 optimization using this DA.com, which converges and goes to completion. This output is stored in the Mol_red1.log file. When I put these .p2n and .log files in a directory and run RED-vIII.pl (with OPT_Calc "Off" and MEPCHR_Calc "On"), I see the following:
* Selected optimization output *
GAUSSIAN
Invalid optimization OUTPUT !
Yes, this is normal. See Tutorial-1.php
Somewhere is this section it is written:
"Finally, the frequency job available in the Gaussian output has to be removed from the QM output (R.E.D. II and III only recognize optimization output, and generates an error message if an optimization job followed by frequency calculations is used as input; this might be changed in R.E.D. IV)."
Simply remove the frequency job from the Mol_red1.log file i.e. from just after the first "Normal termination of Gaussian" set of words to the end of the file. (Frequency jobs are useful to check if the stationary point found is a minimum or a TS).
3) running_RED-III
From: FyD <fyd@q4md-forcefieldtools.org>
Date: 19 Jul 2007
Quoting Karol Kaszuba
I have a couple of questions related to RED-III execution:
1) I have installed resp, gamess(the newest one) and RED-III.
So you have the "ddickick.x", "rungms" and "gamess.00.x" (or "gamess.$n.x" with $n = 00 up to 09) binaries installed in your system.
If your system is a PC, you might also consider using PC-GAMESS since it seems to be really faster (2 times ?). PC-GAMESS can now be interfaced by RED-III. PC-GAMESS is used by RED-III only if at least one of the three GAMES-US binaries is missing. Thus, on our machines, we have GAMESS-US installed in "/usr/local/gamess" and PC-GAMESS in "/usr/local/pcgamess". To use PC-GAMESS I simply remove "/usr/local/gamess" from my "$PATH", and add it again if I want to use it instead of PC-GAMESS ("/usr/local/pcgamess" being always in my "$PATH").
I checked resp and gamess - I put them in my path, I tested them (I am using BASH shell)- they work. In the RED-II manual it is wrote that I have to modify the ".cshrc" file to "run GAMESS, and "RESP program.
R.E.D. checks for binaries using the "which" UNIX command. If the answer to "which resp" is correct, it should be OK. Same rule for the three GAMESS-US binaries.
First question - I am not using .cshrc so I don't have to modify .cshrc to run these programs under RED-III ?
You have to modify your .cshrc file _if_ you use the csh or tcsh shell. Modify the corresponding file for the BASH shell.
2) I modifed three lines in rungms as it is recommended in RED-II manual to:
setenv IRCDATA $SCR/$JOB.irc
setenv INPUT $SCR/$JOB.F05
setenv PUNCH $SCR/$JOB.dat
Correct.
3) I used Ante_RED for p2n file generation - I checked the Ante_RED.log - without any errors, I edited p2n file as it is recommended in RED-III tutorial.
There is a bug in Ante-RED: See the 1st bugfix from the RED home page.
Yes, check in particular:
(i) the atom connectivities which might be a little particular if:
- your starting structure is bad
- you have a metal in your system
(ii) the two columns of atom names
- to select the equivalencing procedure you wish to apply during the fitting procedure (different choices might be made depending on the case you study)
- to select the atom names that will be available in the force field library to the Tripos mol2 file format (two atoms cannot have two identical atom names).
4) I executed RED.pl (as it is recommended in RED-III tutorial) and I noticed an error:
"DO NOT RUN THIS SCRIPT AS ROOT"
so I modified the RED-III.pl script - I deleted lines:
if($< == 0){ print "\n\t\tERROR: DO NOT RUN THIS SCRIPT AS ROOT!\n\n";
if($XRED eq "ON"){ print "\t\tPress Enter to exit.\n\n"; ; }
exit(0);
Yes you can indeed, de-activate this test. However, my understanding is that it is always a _bad_ idea to execute a program as the super-user. Thus, I would not do that. Anyway, it is a good idea to go in the source code and to try to remove tests since you might need to do it in other cases...
- and I tried again - this time "Verifications of variables" passed without any errors but I noticed the second error:
"The initial P2N file can not be found".
R.E.D. needs Mol_red$n.p2n files to run... For instance:
- If $n = 1 (This is a single molecule fitting approach), you only need to provide Mol_red1.p2n (the name being fixed)...
- If $n = 3 (This is a three molecule charge fitting approach), you need to provide three p2n files: Mol_red1.p2n, Mol_red2.p2n and Mol_red3.p2n.
- Now, if you provide Mol_red1.p2n, Mol_red2.p2n and Mol_red4.p2n, R.E.D. will only use the 2 first p2n files since you have a hole in the list of p2n files.
The P2N files and RED-vIII.pl script have to be in the same (working) directory.
I checked everything, I read carefully RED-III tutorial, RED-II manual and I didn't find any clue.
In the tutorial, the RED-vIII.pl scripts have been removed from the data available for download since (i) one has to register to get RED-III and (ii) the tutorial data are free for download.
4) running_RED-III again
From: FyD <fyd@q4md-forcefieldtools.org>
Date: 19 Jul 2007
Quoting Sergey Samsonov
Now the file is found but I get other errors. In case "$OPT_Calc ="On"" I receive:
"ERROR: The Scratch directory is NOT empty"
This means your QM scratch directory is not empty: this has nothing to do with $OPT_Calc ="On" or $OPT_Calc ="Off". Just do a "rm *" in this QM scratch directory. This means you previously run R.E.D., GAMESS or Gaussian and a QM job did not correctly ended. Thus, you still have temporary QM files in your QM scratch directory.
In my .cshrc the QM scratch directory is defined as follows:
setenv SCR /usr/local/0QM_SCR
setenv GAMESS_SCR /usr/local/0QM_SCR
setenv GAUSS_SCRDIR /usr/local/0QM_SCR
The QM scratch file being also set in the "rungms" GAMESS-US script; see "set SCR=/usr/local/0QM_SCR" at the beginning of this file (line ~ 38)
Use "chmod -R 777 /usr/local/0QM_SCR" to get the right permissions in this directory.
I defined an alias in my .cshrc as "alias RM 'rm -rf $SCR/*'".
Then, I clean up the scratch directory each time required using the RM alias instead of the rm command.
In case "$OPT_Calc ="On"" I receive:
"ERROR: The optimization OUTPUT can not be found"
No, here you use $OPT_Calc ="Off"; This means you have to provide the QM ouput (geometry optimization job obtained from a previous standalone execution) as input for R.E.D. The file name for a QM output is once again fixed; it follows similar rules to the P2N file; however with the .log extension instead of the .p2n one; i.e. Mol_red1.log etc...
I tried both own examples and the ones from Tutorial. It seems like I should probably change the names of some files but I cannot understand which ones.
- If $OPT_Calc ="On", you only need to provide P2N file(s).
- If $OPT_Calc ="Off", you need to provide P2N file(s) with its(their) corresponding QM geometry optimization ouput(s). In this case, for each molecule you need to provide a couple of files: Mol_red1.p2n/Mol_red1.log, Mol_red2.p2n/Mol_red2.log etc...
Mol_red$n.p2n/Mol_red$n.log files have to be with the RED-vIII.pl script in the same (working) directory.
5) running_RED-III again and always
From: FyD <fyd@q4md-forcefieldtools.org>
Date: 23 Jul 2007
Quoting Sergey Samsonov
I've been trying again to use RED program (now my GAMESS works fine) and I faced the following error:
The RESP-A1 charges are being derived for molecule 1 ... [ FAILED ]
See the "output(1|2)_m1
For the "RESP-A1" charge model, R.E.D. checks if a charge value equals to 0.000000. If yes, R.E.D. generates an error message (This test is not applied in the case of the "RESP-C2" charge model for instance).
See your RESP output. You have 0.000000 everywhere: Something wrong here:
no. At.no. q(init) q(opt) ivary d(rstr)/dq
1 6 0.000000 0.000000 0 0.005000
2 1 0.000000 0.000000 0 0.000000
3 1 0.000000 0.000000 0 0.000000
4 1 0.000000 0.000000 0 0.000000
5 6 0.000000 0.000000 0 0.005000
6 1 0.000000 0.000000 0 0.000000
7 1 0.000000 0.000000 0 0.000000
8 1 0.000000 0.000000 0 0.000000
You use GAMESS: did you modify the "rungms" script as it is described in the R.E.D.-II manual pages 7-8? It is written:
-------------------------------------
Remarks regarding the use of "GAMESS" with R.E.D.
- The "GAMESS" execution script, named "rungms", has been modified so that "GAMESS" saves its MEP file(s) (".dat" or "punch" file) in its scratch directory ("$SCR"). Thus, R.E.D. finds, and then converts this/these file(s) into the "RESP" format ["espot" file(s)]. See "rungms", around lines 33 for the definition of the "GAMESS" scratch directory, and around lines 73-80 for the modification that has to be carried out:
# "/QM_Disk/QM_SCR" is an example of scratch directory (~ line 33)
set SCR=/QM_Disk/QM_SCR # or whatever is your QM scratch directory
# (A lot of disk space has to remain available for temporary QM files...)
# This is also the scratch directory used by "Gaussian"
# (See the ".cshrc" file provided just above)
# file assignments: # ~ lines 73-80
# all binary files should be put on a node's local disk ($SCR directory),
# both ASCII punch files might be written by NFS to user's permanent disk
set echo
setenv EXTBAS /dev/null
setenv IRCDATA $SCR/$JOB.irc # important modification !!!
setenv INPUT $SCR/$JOB.F05 # important modification !!!
setenv PUNCH $SCR/$JOB.dat # important modification !!!
setenv AOINTS $SCR/$JOB.F08 # important modification !!!
-------------------------------------
6) Reorientation information in R.E.D.
From: FyD <fyd@q4md-forcefieldtools.org>
Date: 24 Jul 2007
Quoting Sergey Samsonov
analyzing the tutorial I've got the question about the re-orientation procedure. What is the emperical rule for estimation who many 'three atoms sets' one need to use for the given molecule?
Using QMRA, you have a single reorientation, and you do NOT control this orientation; or at least not rigorously...
Using RBRA, you fully control the orientation.
Using multi-RBRA, you use several orientations and take into account of the orientational dependence of charge values.
Usually I used only heavy atoms in the set of three atoms.
I select a set of three atoms, let's say atom number X Y Z and to cancel out the orientation effect I do also Z Y X; then, I repeat this 2 orientation procedure several times. See the projectis in R.E.DD.B. you will find many examples.
Probably the increasing number of the three atom sets will increase accuracy but as I understand (I'm not sure I'm right) at the certain point the inclusion of the following sets will be useless. How big could be influence on the error if one uses GAMESS. Another related question is which atoms to choose? In the tutorial it's written that the procedure is arbitrary for heavy atoms...
Well we pick up arbitrary heavy atoms, Yes. Is it shocking ? I do not think so since using QMRA you do not control the orientation at all (worst you beleive you control it)
Now, you can imagine your _own_ strategy. For instance, select common atoms belonging to a family of molecules to induce common orientations whatever the molecule belonging to the family is:
A good example would be to pick up N CA C backbone atoms for all the amino-acids (AA) since they are in all the AA. And then C CA N.
To be continued...
FAQ-II
FAQ-I