[SOLVED]core.io.silent: [ WARNING ] …

Member Site Forums Rosetta 3 Rosetta 3 – Applications [SOLVED]core.io.silent: [ WARNING ] …

Viewing 4 reply threads
  • Author
    Posts
    • #1045
      Anonymous

        Hi List,
        When clustering, I’ve got thousands of the following warning message:
        core.io.silent: [ WARNING ] Number of atoms in pose and silent file disagree! Attempting to continue …
        core.io.silent: [ WARNING ] (in residue 287 natoms_pose=7atm_seqpos 287 natoms_struct=11)
        core.io.silentKilled

        Before clustering, I have gathered several silent files using the full atom mode with the option “silent_struct_type binary”. Is there something to do with the structure type?

        Any clues are appreciated.

      • #6093
        Anonymous

          Was the binary silent file made on the same system as the clustering you are trying to do? The binary file type is not reliably portable between different systems; things like different processor register sizes will alter how the files are interpreted.

        • #6096
          Anonymous

            The warning lines above are the last ones. I mean, the program just stopped at that point without clustering anything. On the other hand, the program seems to work in order when clustering only one of the components silent/binaries files.
            There must be something very wrong. The pdb S_0161_37 bellow is completely different from the native structure which I am modeling only few loops.

            ##################
            protocols.cluster: Adding struc: -166.219
            protocols.cluster: Adding to cluster 4 Cluster_rad: 3
            protocols.cluster: Adding a S_0161_36_1 -166.219
            protocols.cluster: Adding struc: -158.01
            protocols.cluster: Adding to cluster 27 Cluster_rad: 3
            protocols.cluster: Adding a S_0161_37 -158.01
            core.io.silent: [ WARNING ] Number of atoms in pose and silent file disagree! Attempting to continue …
            core.io.silent: [ WARNING ] (in residue 1 natoms_pose=6atm_seqpos 1 natoms_struct=9)
            core.io.silent: [ WARNING ] Number of atoms in pose and silent file disagree! Attempting to continue …
            core.io.silent: [ WARNING ] (in residue 2 natoms_pose=7atm_seqpos 2 natoms_struct=20)
            core.io.silent: [ WARNING ] Number of atoms in pose and silent file disagree! Attempting to continue …
            core.io.silent: [ WARNING ] (in residue 3 natoms_pose=7atm_seqpos 3 natoms_struct=12)
            core.io.silent: [ WARNING ] Number of atoms in pose and silent file disagree! Attempting to continue …
            core.io.silent: [ WARNING ] (in residue 4 natoms_pose=7atm_seqpos 4 natoms_struct=20)
            core.io.silent: [ WARNING ] Number of atoms in pose and silent file disagree! Attempting to continue …
            core.io.silent: [ WARNING ] (in residue 5 natoms_pose=7atm_seqpos 5 natoms_struct=14)
            core.io.silent: [ WARNING ] Number of atoms in pose and silent file disagree! Attempting to continue …
            #################

          • #6098
            Anonymous

              Have a look, please, in the head and tail of the corresponding clustering.log http://ompldr.org/vYW84Yg
              It’s too big to upload it entirely. I put the options I used in the beginning.
              The procedure was as follow:
              1) I run 50 identical process of abnitio loop model;
              2) combined the corresponding silent files;
              3) I cluster the unique all_silent_files.out

              edit: When combining, the program didn’t accept the option “silent_struct_type protein” and warned to use the option binary. Cluster was tried with no special option, just radius 3.0 Ang.

            • #6100
              Anonymous

                Sorry, the script posted was the abnitio loop modeling one which I supposed that was the problem. I think you are right. I didn’t use the option fullatom when clustering – I thought the program would guess since I used such option in the last to steps. Now, I am currently running clustering with fullatom option to see what happens. I’ll let you know the results.
                Thanks for helping.

                edit: fullatom option has solved the problem.

              • #6094
                Anonymous

                  Yep, the binary files were made on the same system and with identical options. Couldn’t be something to do with idealize options?

                • #6095
                  Anonymous

                    I guess it could be. Idealization shouldn’t affect the _number_ of atoms, which is what this error appears to be related to. Design alters the number of atoms. Hopefully Rocco will weigh in, maybe he knows more about this than I do.

                    The warning message is not the location of the crash – that code just prints a warning but does not force an exit. I don’t know how far downstream it is crashing, so it could be unrelated?

                    Another possible cause of an error like this might be mishandling of alternate residue types – for example, if there are loops being modeled, cutpoint variants (with virtual overlap atoms to calculate the loop closure) are used; those extra atoms might be causing the mismatch? Perhaps something like the missing hydrogens due to disulfide bonds could cause the mismatch (that’s not 4 atoms, but it’s the sort of thing to consider). What is going on at the relevant position?

                  • #6097
                    Anonymous

                      The large number of natoms_pose=6 and natoms_pose=7 lead me to believe this may be a centroid/full atom issue. Especially since you’ve tagged this “Loop Modeling”, I’m guessing that for some reason the silent file you created is filled with centroid structures, but the clustering application is expecting full atom structures. … Although you do say that you’ve combined them using full atom, and the individual files seem to work fine. Is it perhaps due to the way you combined them?

                      I think we’re going to need more details. How did you generate the silent files? How did you combine them? How are you running the clustering? Is there anything else that happened to them in the mean time? Representative command lines would be helpful.

                    • #6099
                      Anonymous

                        I just talked James and he agrees that it looks like a centroid/full atom difference, only I got the order the wrong way around (sorry about that). The structs in the silent file are full atom, but the clustering application is expecting to read centroid structures into the pose.

                        The flag file you posted doesn’t look like one used with clustering, but rather the one you used with the abinitio step. Our guess is that in the clustering flag file you left off the “-in::file::fullatom” option. Try following the simple example given in the log file and seeing if it works for you then:

                        cluster -database ~/minirosetta_database -in:file:silent silent.out -in::file::binary_silentfile -in::file::fullatom -native 1a19.pdb

                    Viewing 4 reply threads
                    • You must be logged in to reply to this topic.