Re: [AMBER] MMPBSA.py update install problem

From: Rilei Yu <yulaomao1983.yahoo.com.cn>
Date: Sun, 30 May 2010 11:09:13 +0800 (CST)

Hi Dr. Jason,


Today, I compiled 1.4 AT again, unfortunately, I still could not install it.This time, I did not make parallel.

r.yu.euphoria:~/amber11/src/mmpbsa_py$ nab
usage: nab [-avs] [-c] [-Dstring] [-noassert] [-nodebug] [-o file] [-v] file(s)

r.yu.euphoria:~/amber11/src/mmpbsa_py$ make serial
/bin/bash setup.sh
Python version 2 detected. No file conversion necessary.
/bin/cp MMPBSA.py *pyc /home/r.yu/amber11/bin
/home/r.yu/amber11/bin/nab -o mmpbsa_py_nabnmode mmpbsa_entropy.nab
/bin/mv mmpbsa_py_nabnmode /home/r.yu/amber11/bin


I am afraid i may cannot install this new version. The deadline is coming for me, so at the same time, I also want to  go back to find the problem from the old version. When I finished MMPBSA, the output are always cleaned. Can you tell me how to keep them?
I really want to check them again. Yesterday, I run one subunit with a ligand, it works well. But when I change it to my systerm: two adjacent subunits with a ligand, it fails with minimizatin steps reached 100000.

Best wishes,

Rilei Yu

--- 10年5月29日,周六, Jason Swails <jason.swails.gmail.com> 写道:

发件人: Jason Swails <jason.swails.gmail.com>
主题: Re: [AMBER] MMPBSA.py update install problem
收件人: "AMBER Mailing List" <amber.ambermd.org>
日期: 2010年5月29日,周六,下午9:55

Hello,

Did you do "make parallel" in AmberTools 1.4 before you tried to install MMPBSA.py?  This is a fairly fine point, but MMPBSA.py should be made with a serial version of nab.  When I built mmpbsa_entropy.nab with an MPI-enabled nab, it compiled just fine, but it conflicted with the MPI_WORLD of MMPBSA.py.MPI, but yours doesn't even appear to be compiling properly.

Try re-compiling AmberTools 1.4, but don't make parallel at the end.  Then try to build MMPBSA.py.

On another note, /home/r.yu is your directory, so you typically don't need to use sudo (this is only for /usr/local/ directory).  You may have to use sudo to get rid of some links you created (i.e. nab link in /home/r.yu/amber10/bin/nab), but in general sudo is not needed here.

Good luck!
Jason

--
Jason Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
352-392-4032
On May 29, 2010, at 7:21 AM, Rilei Yu <yulaomao1983.yahoo.com.cn> wrote:
> Dear amber users,
> 
> Very thanks for your former suggestion.
> Tonight, I tried to install mmpbsa.py based on your suggestions. 
> 1) put the path in the .bashrc
> export AMBERHOME=/home/r.yu/amber11
> export PATH=$AMBERHOME/bin\:/home/r.yu/amber10/bin\:$PATH
> 2)
> sudo mv /home/r.yu/amber10/bin/nab /home/r.yu/amber10/bin/nab-1.2
> sudo ln -s /home/r.yu/amber11/bin/nab /home/r.yu/amber10/bin
>  
> 
> 3)
> cd $AMBERHOME/src/mmpbsa_py
> 
> make serial
> 
> make parallel
> 
> r.yu.euphoria:~/amber11/src/mmpbsa_py$ make serial
> /bin/bash setup.sh
> Python version 2 detected. No file conversion necessary.
> /bin/cp MMPBSA.py *pyc /home/r.yu/amber11/bin
> /home/r.yu/amber11/bin/nab -o mmpbsa_py_nabnmode mmpbsa_entropy.nab
> /home/r.yu/amber11/lib/libnab.a(sff.o): In function `mme_timer':
> sff.c:(.text+0x2b40): undefined reference to `lam_mpi_comm_world'
> sff.c:(.text+0x2b4e): undefined reference to `lam_mpi_max'
> sff.c:(.text+0x2b53): undefined reference to `lam_mpi_double'
> sff.c:(.text+0x2bda): undefined reference to `MPI_Allreduce'
> /home/r.yu/amber11/lib/libnab.a(sff.o): In function `mpierror':
> sff.c:(.text+0x86c6): undefined reference to `lam_mpi_comm_world'
> sff.c:(.text+0x86cc): undefined reference to `lam_mpi_min'
> sff.c:(.text+0x86df): undefined reference to `lam_mpi_int'
> sff.c:(.text+0x86e9): undefined reference to `MPI_Allreduce'
> sff.c:(.text+0x8701): undefined reference to `MPI_Finalize'
> /home/r.yu/amber11/lib/libnab.a(sff.o): In function `mpiinit':
> sff.c:(.text+0xca2c): undefined reference to `MPI_Init'
> sff.c:(.text+0xca4d): undefined reference to `lam_mpi_comm_world'
> sff.c:(.text+0xca52): undefined reference to `MPI_Comm_rank'
> sff.c:(.text+0xca62): undefined reference to `lam_mpi_comm_world'
> sff.c:(.text+0xca67): undefined reference to `MPI_Comm_size'
> 
> So I went back to re-install the tool.1.4 again, and found the same problem. Can you give me more suggestions.
> 
> I am really appreciated for your suggestions again!
> Best wishes,
> 
> 
> Rilei Yu
> 
> --- 10年5月29日,周六, Jason Swails <jason.swails.gmail.com> 写道:
> 
> 发件人: Jason Swails <jason.swails.gmail.com>
> 主题: Re: [AMBER] MMPBSA.py update install problem
> 收件人: "AMBER Mailing List" <amber.ambermd.org>
> 日期: 2010年5月29日,周六,上午4:38
> 
> On Fri, May 28, 2010 at 1:20 PM, Bill Ross <ross.cgl.ucsf.edu> wrote:
> 
>>> When you install AmberTools 1.4 you will not need to uninstall AmberTools
>>> 1.2 or Amber 10. Instead, just place the path for Amber 11 (export
>>> AMBERHOME=/usr/local/amber11) before defining the path for Amber 10
>>> (AMBERHOME=/usr/local/amber10) in your .bashrc file. This will set up
>> your
>>> environment such that the computer will look for the executable files in
>>> Amber 11 *before* the executable files in Amber 10, which is what you
>> want.
>> 
>> It doesn't work this way - your environment has only one AMBERHOME
>> variable, it is not like a path where directories are indexed
>> sequentially.
>> 
> 
> Yes. One at a time.  To install amber10, though, AMBERHOME needs to point to
> /usr/local/amber10 temporarily (while the install is taking place).  Then it
> should point to /usr/local/amber11 to install AmberTools 1.4, and it should
> remain pointing to /usr/local/amber11 after it's done (since AmberTools
> makes the most use of AMBERHOME; indeed probably the only use of it).
> However, in order for MMPBSA.py to operate successfully, both
> /usr/local/amber10/bin and /usr/local/amber11/bin must be in the PATH, with
> /usr/local/amber11 coming first so AmberTools-1.4 binaries are used before
> AmberTools-1.2 that reside in /usr/local/amber10.  This is what Bill meant,
> I believe. (Though defining AMBERHOME a second time in the .bashrc does
> overwrite the first definition).
> 
> For installing MMPBSA.py, though, you should probably create a symbolic link
> from nab in /usr/local/amber11/bin to the directory /usr/local/amber10/bin
> in order to build the nmode program with the proper nab compiler.  You can
> do this as follows:
> 
> sudo mv /usr/local/amber10/bin/nab /usr/local/amber10/bin/nab-1.2
> sudo ln -s /usr/local/amber11/bin/nab /usr/local/amber10/bin
> 
> I used sudo because typically users don't have write permissions in
> /usr/local, but if you changed the ownership of /usr/local/amber10 and 11,
> then you don't need the sudo (but if you don't need the sudo, you generally
> know you don't...)
> 
> What this does is to move nab from AT 1.2 "out of the way" so that a soft
> link can be put in /usr/local/amber11/bin, so that MMPBSA.py installation
> will use that.
> 
> Ideally what should be in your .bashrc should look something like this:
> 
> export AMBERHOME=/usr/local/amber11
> export PATH=$AMBERHOME/bin\:/usr/local/amber10/bin\:$PATH
> 
> Hope this helps,
> Jason
> 
> 
>> Bill
>> 
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>> 
> 
> 
> 
> -- 
> Jason M. Swails
> Quantum Theory Project,
> University of Florida
> Ph.D. Graduate Student
> 352-392-4032
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
> 
> 
> 
> 
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat May 29 2010 - 20:30:03 PDT
Custom Search