Generic Problems and Workarounds

Release 2.0

There are two bugs with the application installation script, CDROM/Midas-2.0/apps/install. A replacement installation script is available.
1. Motif/OpenGL versions of midas (e.g. DEC Alpha) need an applications default file installed before they can run successfully. Either use the replacement script above or do the following after the installation is done:
```
cp /usr/local/midas/resource/midas/Midas.ad /usr/lib/X11/app-defaults/Midas
```
2. If midas is first installed to run from the CDROM and then reinstalled on a disk, the installation fails because the /usr/local/midas symbolic link is not removed. Either use the replacement script above or remove /usr/local/midas first.
Some versions of the MidasPlus User's Manual do not describe how to use MidasPlus as a ``PDB viewer'' from applications such as Mosaic. The following paragraph explains what needs to be done:
The -D option causes Midas to run in the foreground instead of automatically backgrounding itself. This option is necessary if Midas is to be used as a MIME viewer. For example, adding the lines
```
chemical/x-pdb; midas -D %s
image/x-pdb; midas -D %s
```
to one's $HOME/.mailcap would allow both World Wide Web viewers, such as Mosaic, and MediaMail to view PDB files. Also see the documentation for ``mailcap.''
The -D options is not needed for the OpenGL nor the NeXT versions of Midas.
It is not possible to switch from one session to another without exiting MidasPlus.
On color hardcopies generated with the copy command, bonds may cross in front of bonds that they should cross behind because bonds are not depth-sorted when generating 2-D PostScript.
Picking does not work well when MidasPlus is in ``stereo walleye/crosseye'' mode. Users should pick on the right eye image only.
It is not possible to have the atoms specified in the watch command checked against a subset of the remaining atoms.
There can be no more than 16 distance, angle or rotation monitors.
The documentation for the bs utility fails to mention explicitly that when MidasPlus is used to generate an input PDB file for bs, that the pdbrun command must be used to do so, and moreover that the ``conect'' keyword argument must be given as well, like so:
pdbrun conect cat > bs_input_file_name
alternatively, the bs output could be directly printed from MidasPlus, like so:
pdbrun conect bs bs_arguments | lpr -Pprinter_name
In order to get MidasMenu to work correctly, you need to run the following command after installation:
chmod o+r /usr/local/midas/resource/command.mmc
The installation guide does not explicitly discuss installation of the MidasPlus manual pages. All the MidasPlus manual pages are found in /usr/local/midas/man. One option is to copy them into the directory tree that the man program searches by default. We recommend against this option since the MidasPlus deinstall script (used when upgrading or removing MidasPlus) won't delete the duplicated files. Instead, it is better if MidasPlus users can set the MANPATH environment variable so that the man program can find the MidasPlus manual pages. The proper setting for the MANPATH variable varies from architecture to architecture, as follows:

SGI workstations
setenv MANPATH /usr/share/catman:/usr/share/man:/usr/catman:/usr/man:/usr/local/midas/man
If the above setenv command is placed in the file /etc/cshrc, it will take effect for every user of the system.
DEC Alpha workstations
setenv MANPATH /usr/share/man:/usr/local/man:/usr/local/midas/man
NeXT workstations
setenv MANPATH /usr/man:/usr/local/midas/man
The man page for the neon command contains an error. In the "EXAMPLES" section on page 133, example #2 should be

neon -F -ofigure.i
and example #3 should be

conic -cparm -F -ofigure.i site
That is, use a capital ``-F'' flag instead of lowercase ``-f'' flag.
The table of Pdbrun keywords, on page 37, is missing the CNAME keyword. CNAME's are found before each model and give the definition of a symbolic color name.
Conic sometimes gets the color wrong when spheres are clipped. This problem will be fixed in the next release.
The density delegate doesn't parse X-PLOR .map files correctly; it needs a space between each number. Until the next release, fix it by doing

sed "s/-0\./ -./g" <original.map>fixed.map
There is a bug regarding the saving of PDB CONECT records. If you work strictly with proteins or nucleic acids having standard residues, or if you have templates for all of your residues in a model, there is no problem. For situations other than these, session files may not preserve the indicated connectivity properly. To get around this problem, supply templates for non-standard residues (see gentpl(1)). This problem also affects the generation of CONECT records for the pdbrun command; the workaround here is to use the "conect" [sic] keyword when invoking pdbrun, i.e. "pdbrun conect". If these two workarounds prove insufficient, please send e-mail to midas-bugs@cgl.ucsf.edu.
Due to the way that MidasPlus 2.0 positions newly opened models, ambershow and noeshow may not orient the contraints on the model structure correctly. This problem can be fixed simply by changing one line in the ambershow and noeshow scripts. The scripts can be found in /usr/local/midas/lib/midas. The line in the each script that is currently:
```
print "open obj $tempf\n";
```
should be changed to:
```
print "open original obj $tempf\n";
```
In the fix, reverse, and fixreverse commands, if the rotation number is followed by any whitespace, the command will fail.
The midas command pdbrun can cause a core dump if a bond angle is active.
Picking of rotated atoms only works on NeXT workstations. A workaround is to fix the bond rotation before picking.
Some match commands that do not specify atom names (e.g. "match #0 #1") and that involve models with non-standard residues (or atom names) may not work as expected. This is because when atoms are not explicitly specified in the command, each residue's atoms are matched in "connectivity" order. If Midas has to figure out the connectivity of a residue on its own (because there is no standard template), then the connectivity order of two residues may vary, even though they are composed of the same atoms. The workarounds available are to either explicitly specify atom names in the match command (you may want to use an alias if it is a long list), or to generate templates for the non-standard residues as described in section 3.6 of the manual (``More on Connectivity and Templates'').
The "group" file describing threonine places the atoms in the incorrect chirality. The workaround is to edit the file /usr/local/midas/resource/groups/THR and change the last number on the last line from 62.7 to -62.7.
Neon may fail to fully render a complex scene in high detail (i.e. parts may be missing). This is because such a scene decription may require neon to communicate more than 100,000 PDB ATOM records to conic for imaging, which overflows the 5 digit atom serial number field of a PDB record (causing the field to be filled with asterisks). This problem can be worked around by issuing the following MIDAS command instead of neon:
```
pdbrun conect surface /usr/local/midas/bin/neon | sed 's/\\*\\*\\*\\*\\*/00000/' | /usr/local/midas/bin/conic
```
You may be able the omit the "/usr/local/midas/bin/"s in the above and have it still work. If you would normally supply options to the neon command, add them to the end of the above command.
Neon will fail if the number of input atoms from Midas exceeds 32767. This is fixed in the next release. There is a workaround for situations where system being modelled in Midas has more than 32767 atoms, but less than that number are in the desired view; undisplay the non-visible atoms before running neon, with:
```
show /vis za<5
```
The above command will undisplay any atoms more than 5 angstroms from atoms currently visible in the view window. It is desirable to keep the atoms within 5 angstroms so that neon will correctly image bonds that go off the side of the view. Also, see the previous bullet item for a related problem with large systems.
The lowest sphere density that neon handles correctly is 2. A bug causes '1' to be treated as if it were '9'. Don't ask. Fixed in the next release.
Esp may produce spurious results for terminal nucleic acid residues. This problem can be corrected by editing the charges data file /usr/local/midas/resource/charges.esp. At the top of this file is the data for the various nucleic acid residues, one atom per line. The residues are listed in order: G, C, A, T, (then others). You should change the last two lines of each nucleic acid residue (the lines for atoms "OH" and "HO") by inserting a "#" at the start of each line (which changes the line to a comment).
Aliases longer than 200 characters can cause MidasPlus to crash. This is fixed in the next release. You may want to break really long aliases into multiple somewhat smaller aliases to work around this problem.
In session files, the colors associated with atoms are written in the order (bond, label, vdw) and read in the order (bond, vdw, label). Fixed in the next release.
"color byatom" doesn't color hydrogens which have leading digits in their names. Fixed in the next release.
An object's displayed-ness is not saved in session files.

[end of list]

midas-bugs@cgl.ucsf.edu / MidasPlus Generic Problems / October 1994