Activity for Andreas Vitalis
-
Andreas Vitalis posted a comment on discussion Help
Hi Matt, that is a curious issue. From https://gcc.gnu.org/onlinedocs/gfortran/DATE_005fAND_005fTIME.html , gfortran simply expects "default" character values of certain minimum lengths. The definitions of the first 3 variables passed to DATE_AND_TIME are, throughout (this affects 4 files): character (len=12) rc(3) However, the fourth argument, specifically in utilities.f90 but not elsewhere, is specified as: integer(KIND=8) dts(8) These KIND specifications are not really recommended (but haven't...
-
Version 5 released!
-
Hi, the mention of the file fftw3_SUN.f03 implies that you are trying to use the Oracle (Sun) compiler. Is this correct? If no: You are accidentally passing "-DORACLE_FORTRAN" in your Makefile (just grep for this string). I don't know where that would be coming from, though. Removing it from the Makefile and recompiling everything should fix it (make clean; make all install). If yes, you have a few options: 1. Use the GNU compilers instead (rerun configure, and recompile) 2. Install FFTW3 using the...
-
Hi Irem, harmonic angular restraints on arbitrary atoms are not supported because they create force discontinuities when the angle approaches 180 deg. , which would take some extra mechanisms to protect from. This is not an issue for standard topological bonds that contain no angles near 180 deg. Unsupported residues or small molecules can contain sp carbons of course, and I don't recall from top of my head what happens to those in Cartesian dynamics. It's probably possible to mimic an angle restraint...
-
I am reposting the message I sent you for reference here: The "make campari" error is, as you diagnosed, because there is no file called "Makefile" by default. The second error is probably because you have not created the lib/bin subfolders beforehand. To avoid linking errors downstream, I would follow the official instructions (which also contain a sample Makefile.local for Linux systems with Gnu compilers). I am going to assume for simplicity that you work in the folder /software/campari With configure:...
-
Hi, apologies for the extraordinary delay. Almost certainly you solved this but for reference: Indeed, the XDR library has issues with Ubuntu 22 and newer. Some components of RPC were migrated out, which leads to the new requirements to: 1) Install the extra package (as you write) 2) Add "-I/usr/include/ntirpc" to the C compiler flags (as you write) 3) Add "-ltirpc" to the linker flags Cheers, Andreas
-
I am sorry for the inordinate delay, but in case this question still matters to you or others: I believe the error is because there is a limit parameter that is not correctly set. Note that this feature is largely obsolete (as the documentation states). I can reproduce a seg-fault if in "mod_grid.f90" I have: integer MAXGRID parameter (MAXGRID=900) But it goes away if I recompile with: integer MAXGRID parameter (MAXGRID=1300) The grids are simply setting up a biased sampler that is meant to prevent...
-
Hi Feng, I am sorry I completely missed this. "install" is part of coreutils, so should pretty much always be there on Linux. It might be that the lines right before failed due to permissions. This could happen if you used the route through Makefile.manual and didn't redefine DESTDIRBIN and DESTDIRLIB there. The documentation doesn't specifically cover this case (running "make install" in the manual route). If this happens with the configure script, it should throw an error earlier. We'll look into...
-
Hi Jihye, not sure I can correlate the keywords exactly with your question (FMCSC_XYZOUT, FMCSC_NRSTEPS and FMCSC_EQUIL give me 10M production steps and 500 written snapshots for a single run) but the problem might be that some specific amino acids, specifically SER/THR/TYR, fail. I am assuming you are running blocked amino acids (aka , dipeptides) so something like ACE-SER-NME. Crashes might happen because you didn't use FMCSC_FUDGE_DYN 1 but a time step of 4fs. The CCOH-dihedral angle is a very...
-
Hi Yifan, this issue is an odd one. We have recently gained access to our latest HPC architecture, which have 2x64-core AMD Epyc processors on them, and I did not encounter any obvious problems. When you try V4 with gfortran, make sure to disable the Fortran08 MPI module (pass "-DDISABLE_MPIF08"). There is a currently unresolved bug in gfortran with the module, at least with the MPICH-based module I have available on my HPC system (note that V3 is not using the Fortran08 MPI module, so this problem...
-
Version 4 released!
-
Hi Yifan, that is a challenging question to answer since I don't have access to such a machine. Generally speaking, you can completely remove PME form the code by simply not linking to FFTW during compilation (so do not specify -DLINK_FFTW). I don't know what exact simualtions you run, but using PME would only be relevant for explicit solvent simulations. I think there is only one tutorial that actually uses PME. Whenever keyword FMCSC_LREL_MD is not 2, no Ewald routines are used in any way. If you...
-
Sorry about the delay: At the moment, you would have to use it as an unsupported. Charges you could get from CHARMM via CgenFF or by analogy. Depending on your plans, though, it could be rather tricky to parameterize. The energy scales for true divalent ions get very large, which also very negatively impacts sampling (effective time scales). In ABSINTH (and in line with most force fields), we have been avoiding divalent ions because of this. That said, I know that Martin Fossat in the Pappu lab was...
-
Hi Gianluca, sorry for the delay: 1) Citations: no, not explicitly for docking. The polyQ work I mentioned is here: https://doi.org/10.1016/j.jmb.2008.09.026 https://doi.org/10.1016/j.bpj.2009.05.003 The other works were specifically interested in characterizing or predicting interfaces but rather simulated known complexes. Marco Bacci has done the most with that on our end recently. 2) The offset is on a parameter that is the reference free energy of solvation for a solvation group. In ABSINTH,...
-
Hi Gianluca, 1) We have simulated several protein complexes in the past. In the earliest days, we were interested in the binding affinity of globules of polyQ. That is a case where we did a comparison of affinity in a rigid approximation vs flexible, and we found that flexibility increases affinity (to be expected, though, as the interfaces were not "optimized" for binding). So both (rigid/flexible) are possible. Most other protein complexes were treated with some amount of flexibility. Marco may...
-
Hi Anne, first apologies for the delay. Cyclic peptides are a tricky issue since most of CAMPARI's functionality (the MC part in particular) is built on the logic that torsion angles are the degrees of freedom. With a cyclic constraint, every backbone move in torsional space has to be of a concerted rotation type. There are two possible solutions: o You can work in Cartesian space (see FMCSC_CARTINT) and treat the molecule has a single, unsupported "residue". This requires a lot of patches should...
-
Hi Chris, this is way outdated, I apologize. For restarting runs, look at FMCSC_RESTART, for analyzing data on the fly, just analyze the trajectories being written (this is usually safe as read access does not conflict with them being open for writing) using FMCSC_PDBANALYZE . If you want more control have a look at FMCSC_FRAMESFILE. Cheers, Andreas
-
Hi Chris, you've got several issues. First, in Makefile, you should change: COMPDEFAULTS=${INTELDEFAULTS} from COMPDEFAULTS=${GNUDEFAULTS} Second, you need to worry about where you are going to get your dependencies from, in particular NetCDF. For example, if you use only system packages, your Makefile.local would include lines like these: EXTRA_LIBS=-lnetcdff -lnetcdf -lfftw3_threads -lfftw3 -llapack MPIEXTRA_LIBS=-lnetcdff -lnetcdf -lfftw3_threads -lfftw3 -llapack HSLEXTRALIBS=-lblas However, for...
-
Hi, is there anything more to this error? What were the few lines before? Please do: "make clean", the "./make_dependencies.sh ." and then "make campari_threads" or "make campari". Your architecture is definitely not a problem. We use Intel compilers on Ubuntu ourselves. I should be able to help you more with that information available. Cheers, Andreas
-
Hi Steffen, sorry I overlooked this. Out of curiosity: what type of benchmark were you looking at for the speed comparison? OpenMP/no OpenMP? It's rare that on a standard CPU, you would get such large differences. One reason could be that gfortran failed to pull the right instructions sets through "march=native". I don't have a Mac lying around, so I cannot test easily.
-
I would probably keep the neutralizing ones as excess, which is more likely to be similar to reality (ionic atmospheres and all).
-
Hi Jonathan, I hope this is answering your question: The sequence file always has to be explicit, i.e.,, all residues, which includes ions, have to be listed, one per line. So it depends on your volume and the numbers you put. Keep in mind that in a droplet the formal volume and the actual volume are not exactly the same because the wall is soft. E.g,, 150mM in a 50A radius droplet using the formal volume is between 47 and 48 cat- and anions each. There is currently no way to shorten the sequence...
-
Hi Dillion, Did you get N_000.log or N_001.log? Once CAMPARI starts, almost all errors are printed there. So if you have a problem with the key-file, CAMPARI would throw an error and then trigger MPI_ABORT. If there aren't too many MPI ranks I usually do st like: tail N_*.log The error usually shows up only in the rank to first encounter it (and triggers the destruction of the MPI universe), which can be random. If that is not the problem, let me know! Andreas
-
Hi Dillion, Did you get N_000.log or N_001.log? Once CAMPARI starts, almost all errors are printed there. So if you have a problem with the key-file, CAMPARI would throw and error and than trigger MPI_ABORT. If there aren't too many MPI ranks I usually do st like: tail N_*.log The error usually shows up only in the rank to first encounter it (and triggers the destruction of the MPI universe), which can be random. If that is not the problem, let me know! Andreas
-
Hi Maria, in general, yes. If you can afford it, you could also set up some comparison runs with abs3.4_charmm36.prm. These parameters are meant to correct some weaknesses identified over the years with the abs3.2... parameters for free energies of solvation of inorganic ions and selected Lennard-Jones parameters. If you run into unexpected behavior, please let us know! Andreas
-
Hi Tere, I am still not entirely sure what the error message I am supposed to help you with is. In general, I agree of course - I will see if anyone has a box with a CentOS 6.9 installation around. Once I have done so, I'll get back to you. Cheers, Andreas
-
Hi all, Marco and I have updated the web docu to be more specific about both HSL and NetCDF: http://campari.sourceforge.net/V3/install.html http://campari.sourceforge.net/V3/download.html Unfortunately the web docu is still very slow at the time of writing, which I assume is due to the data center migration of SF: https://twitter.com/sfnet_ops Cheers, Andreas
-
I think you had " -std=f2008" before, so can you also repost the error you get with that? The current error is due to INT64 being a 2008 feature. One thing you can try in addition is to pass "-DORACLE_FORTRAN" just for the compilation of graph_algorithms.f90 and thread_utils.f90. In other words, try this: gfortran -DORACLE_FORTRAN -DDISABLE_FLOAT -DLINK_XDR -DLINK_FFTW -DLINK_NETCDF -I/data2/teresa/campari/local_libs/local/include -I/data2/teresa/campari_abs/lib/x86_64/ -c -cpp -O3 -fall-intrinsics...
-
Hi Steffen, SEP is -1. All phosphorylated residues are in what is probably the most populated species, i.e., the hydrogenphosphate ester (-). I wouldn't recommend trying to simulate the naked terminal phopshate (2-) in any case (not that that is supported at the moment). I am not entirely sure that answers your question, though. Andreas
-
Well, two things come to mind: 1) You may want to consider keeping the backbone charges from OPLS-AA/L. In both CHARMM and OPLS, the NHCHCO scaffold is always net neutral and does not vary with sidechain. For the rest, you'll just need to define new charge types. The phosphate groups are parameterized to be singly negatively charged, so they carry one OH. CAMPARI will complain if the charges do not add up correctly, so this part is usually relatively easy to debug. 2) Bonded paramteters are needed...
-
Hi Steffen, OPLS does not have native support for phosphorylated residues, so there are no bonded parameters and no partial charges. The easiest is to use abs3.2_charmm.prm or abs3.2_charmm36.prm instead (the charge sets between OPLS and CHARMM are generally quite similar). If you want to compare to OPLS runs that have lready been completed, you can also specifically transfer over the parameters from abs3.2_charmm36.prm to merge into a custom copy of abs3.2_opls.prm, but this is a bit more work....
-
Hi, in your "Makefile.local", simply do not pass "-DLINK_HSL" and it should not try to compile HSL at all (in the docu, this option is included for "FFLAGS"). If that is not enough (it should be, though), you can also set the flag "USE_INTERNAL_HSL" to 0 in "Makefile". I'll ask my colleague about the default location for common90.f90, and we'll get back to you on that. Hope that helps (for now), Andreas
-
Hi Steffen, I just uploaded the code. Best, Andreas
-
Version 3 released!
-
Hi Steffen, phosphorylated S, T, and Y are fully supported residues in version 3. I'll upload a very-late-beta of version 3 very soon. Andreas
-
Hi Carlos, with FMCSC_MPIAVG you can simulate identical copies of a system in parallel...
-
Hi Hannah, some more information would be needed, i.e., what operating system are...
-
Hi Ben, could you try to modify "-std=f2003" in the Makefile to "-std=f2008" and...
-
Hm, so you used FMCSC_XYZ_FORCEBOX during the CAMPARI run? Normally, CAMPARI will...
-
You have a couple of typos in the relevant keywords: FMSCSC_RANDOMIZE 1 # generate...
-
Hi Ramon, the LJ parameters are not printed this way because they are actually pairwise...
-
Hi Ramon, the LJ parameters are not printed this way because they are actually pairwise...
-
1) You should be able to - relatively easily - produce a .tpr file from GROMACS....
-
Hi, that makes it clear. The unsupported options are in fact GNU options. In the...
-
Good question. It is definitely not needed to go to -O2. What flags does it show...
-
Hi, this is almost always a result of aggressive compiler optimization (e.g., computing...
-
Hi Carlos, At first glance, I would say that it looks like you have built NetCDF...
-
Hi Logan, yes, there are 4 major ways of acheiving secondary structure bias. 1) FMCSC_SC_ZSEC...
-
Hi again, I just tested an MPI compile from the latest zip on here on Ubuntu 14 with...
-
Hi Logan, so sorry I overlooked this. Your first understanding is correct. If you...
-
Hi, the obsolescent warnings you can ignore. FYI, V3 will not use the old mpi.h include...
-
Yes, that should work.
-
Hi Shubham, unfortunately there is no direct option for that. Velocities are either...
-
Could you please tell me which change made it work? That way I know whether I should...
-
Hi Shubham, segmentation faults can be anything. From the trace, I would think that...
-
Dear Shubham, yes it can. The main issue is the choice of degrees of freedom (and...
-
As I said in one of the replies above, before addressing this question, one should...
-
As I said in one of the replies above, before addressing this question, one should...
-
As I said in one of the replies above, before addressing this question, one should...
-
I would say there are two effects here: 1) In a polymer and irrespective of solvent...
-
Hi Allison, this is a twofold question but the answers may be related: A) Difficulty...
-
With SAV analysis, you could raise the thickness of the solvation layer (FMCSC_SAVPROBE)...
-
A) Sorry, you are absolutely right: you do have to edit the restart file to move...
-
Just to add to Maxim's answer: If you do not have a PDB, the initial sturcture of...
-
Dear Guang, the un-rexing of replica exchange trajectories is possible in trajectory...
-
See http://campari.sourceforge.net/inputfiles.html#FMCSC_REFILE The target alpha-content...
-
Hi Guang, in terms of input, your understanding is correct. Functionally, the main...
-
Hi Guang, in terms of input, your understanding is correct. Functionally, the main...
-
Hi Maxim, with multiple molecules, you have to worry about the boundary regardless....
-
Hi Ramon, there should be two easy solutions: First, just to go to a large enough...
-
Hi Maxim, did this comment change substantially? It was on my list to reply to it...
-
Hi Sharon, the error you are getting for the route using system libraries indicates...
-
Hi Sharon, for most libraries on a standard Linux kernel, you can use the system-installed...
-
I do not think that there is ever the case of a clear "best" value for the swap frequency....
-
Hi Ramon, we do know that changing the dielectric often has a very small impact (tried...
-
Hi Ramon, did this reply work for you? I would also like to alert you that I just...
-
I just uploaded an archive with a number of bug fixes. In the process, I have also...
-
Another collection of bug fixes
-
Hi Maxim, for the move set adjustments, this is mostly what I meant. For the concerted...
-
The operational defintion of the cuboid is indeed the vector you give (all coordinates...
-
Dear Maxim, thanks! It is great to know that CAMPARI is useful to you. I took the...
-
Dear Ramon, this is not high on our lists. As you hint at, for an implicit model...
-
Dear Ramon, this is not high on our lists. As you hint at, for an implicit model...
-
Hi Ramon, keyword FMCSC_RETRACE turns on the printout of the REX trace file (N_000_REXTRACE.dat)....
-
The difference in tutorial 3 is that there are explicit counterions, which is why...
-
The difference in tutorial 3 is that there are explicit counterions, which is why...
-
Hi Ramon, so sorry that I overlooked this post. I hope you have found answers in...
-
Collection of bug fixes
-
Hi Ramon, I believe the key to your question is keyword FMCSC_ALIGN. It controls...
-
a) The computational complexity of constraints vs. restraints can matter substantially...
-
Hi Ramon, I try to answer one-by-one but not in chronological order: a) You always...
-
Hi Ramon, I try to answer one-by-one but in chronological order: a) You always need...
-
Bug fix to replica exchange
-
Bug fix to replica exchange
-
Bug fix to replica exchange
-
Bug fix to replica exchange
-
Hi Daniel, thanks for your interest. FMCSC_SC_BONDED_B 0.0 # Scaling factor for bonded...
-
Minor fixes to CAMPARI V2
-
Dear Ramon, RMSD restraint potentials will probably be included at some point. Since...
-
Dear Ramon, 1) Unless there are specific questions you have, the reference http://pubs.acs.org/doi/abs/10.1021/ct200744s...
1 >