[COOT] Bus error when accessing Ramachandran Plot or Kleywegt Plot

2024-07-24 Thread Markus Meier
rn({mask=[]}) = 0
rt_sigprocmask(SIG_BLOCK, NULL, [], 8)  = 0
rt_sigprocmask(SIG_BLOCK, [INT TERM CHLD], [], 8) = 0
clone(child_stack=NULL, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, 
child_tidptr=0x14562dbb6450) = 1591821
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
rt_sigprocmask(SIG_BLOCK, [CHLD], [], 8) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
rt_sigprocmask(SIG_BLOCK, [CHLD], [], 8) = 0
rt_sigaction(SIGINT, {sa_handler=0x5559026814d0, sa_mask=[], 
sa_flags=SA_RESTORER, sa_restorer=0x14562dc2d3d0}, {sa_handler=SIG_DFL, 
sa_mask=[], sa_flags=SA_RESTORER, sa_restorer=0x14562dc2d3d0}, 8) = 0
wait4(-1, Guile 1.8.8
Copyright (c) 1995, 1996, 1997, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 
2008 Free Software Foundation
Guile may be distributed under the terms of the GNU General Public Licence;
certain other uses are permitted as well.  For details, see the file
`COPYING', which is included in the Guile distribution.
There is no warranty, to the extent permitted by law.
[{WIFEXITED(s) && WEXITSTATUS(s) == 0}], 0, NULL) = 1591821
rt_sigaction(SIGINT, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=SA_RESTORER, 
sa_restorer=0x14562dc2d3d0}, {sa_handler=0x5559026814d0, sa_mask=[], 
sa_flags=SA_RESTORER, sa_restorer=0x14562dc2d3d0}, 8) = 0
ioctl(2, TIOCGWINSZ, {ws_row=26, ws_col=137, ws_xpixel=0, ws_ypixel=0}) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
--- SIGCHLD {si_signo=SIGCHLD, si_code=CLD_EXITED, si_pid=1591821, si_uid=1002, 
si_status=0, si_utime=0, si_stime=0} ---
wait4(-1, 0x7ffc699af4d0, WNOHANG, NULL) = -1 ECHILD (No child processes)
rt_sigreturn({mask=[]}) = 0
write(1, "catching the crash log:\n", 24catching the crash log:
) = 24
rt_sigprocmask(SIG_BLOCK, NULL, [], 8)  = 0
rt_sigprocmask(SIG_BLOCK, [INT TERM CHLD], [], 8) = 0
clone(child_stack=NULL, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, 
child_tidptr=0x14562dbb6450) = 1591822
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
rt_sigprocmask(SIG_BLOCK, [CHLD], [], 8) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
rt_sigprocmask(SIG_BLOCK, [CHLD], [], 8) = 0
rt_sigaction(SIGINT, {sa_handler=0x5559026814d0, sa_mask=[], 
sa_flags=SA_RESTORER, sa_restorer=0x14562dc2d3d0}, {sa_handler=SIG_DFL, 
sa_mask=[], sa_flags=SA_RESTORER, sa_restorer=0x14562dc2d3d0}, 8) = 0
wait4(-1, Gtk-Message: 12:45:07.416: Failed to load module "canberra-gtk-module"
coot-exe: 
"/software/share/coot-0.9.8.95/coot-Linux-GenuineIntel---gtk2-python/libexec/coot-bin"
/usr/bin/ls
-rwxrwxr-x 1 admin admin 302519496 Jul 23 13:29 
/software/share/coot-0.9.8.95/coot-Linux-GenuineIntel---gtk2-python/libexec/coot-bin
coot-version:
/software/share/coot-0.9.8.95/coot-Linux-GenuineIntel---gtk2-python/libexec/coot-bin
Builder_info: built by:  build-host: luna on: Tue 23 Jul 1316.55 CDT 2024
Enabled: Enhanced-ligand-tools C++-11 Threads Boost-based-thread-pool Goocanvas 
GSL SQLite3 LibCurl
Binary type: Linux-GenuineIntel---python-gtk2
git commit: edd12aa553da94a40dfb5152cae6f25991654815
Builder_info: built by:  build-host: luna on: Tue 23 Jul 1316.55 CDT 2024
[with python 2.7.18 embedded]
[with guile 1.8.8 embedded]
0.9.8.95 (revision-count 10979)
platform:
/usr/bin/uname
Linux luna 5.10.216-gentoo #2 SMP Tue Jul 2 12:13:56 CDT 2024 x86_64 Intel(R) 
Xeon(R) CPU E5-2680 v3 @ 2.50GHz GenuineIntel GNU/Linux
core: #f
No core file found.  No debugging
[{WIFEXITED(s) && WEXITSTATUS(s) == 0}], 0, NULL) = 1591822
rt_sigaction(SIGINT, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=SA_RESTORER, 
sa_restorer=0x14562dc2d3d0}, {sa_handler=0x5559026814d0, sa_mask=[], 
sa_flags=SA_RESTORER, sa_restorer=0x14562dc2d3d0}, 8) = 0
ioctl(2, TIOCGWINSZ, {ws_row=26, ws_col=137, ws_xpixel=0, ws_ypixel=0}) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
--- SIGCHLD {si_signo=SIGCHLD, si_code=CLD_EXITED, si_pid=1591822, si_uid=1002, 
si_status=0, si_utime=4 /* 0.04 s */, si_stime=0} ---
wait4(-1, 0x7ffc699af350, WNOHANG, NULL) = -1 ECHILD (No child processes)
rt_sigreturn({mask=[]}) = 0
rt_sigprocmask(SIG_BLOCK, [CHLD], [], 8) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
exit_group(135) = ?
+++ exited with 135 +++


--
Markus Meier, Ph.D.
Research Associate
University of Manitoba
Department of Chemistry
144 Dysart Road
Winnipeg, MB, R3T 2N2, Canada
Phone:  +1 204 474 7172
E-mail: markus.me...@umanitoba.ca



To unsubscribe from the COOT list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=COOT=1

This message was issued to members of www.jiscmail.ac.uk/COOT, a mailing list 
hosted by www.jiscmail.ac.uk, terms & conditions are available at 
https://www.jiscmail.ac.uk/policyandsecurity/

20-coot.txt.bz2
Description: BZip2 compressed data


[COOT] The carbohydrate module in Coot 0.9.8.95 requests model-level-2-FUC-ALPHA1-6-NAG.tab

2024-07-24 Thread Markus Meier
ype_filter_general() for ALPHA1-2
link_type_filter_general() checking type ALPHA1-2 and returns ""
link_type_filter_general() for ALPHA1-2
link_type_filter_general() checking type ALPHA1-2 and returns ""
INFO:: replace_coords: 24 atoms updated.
find_rooted_tree returns tree:
 ASN [spec: 1 "A" 133 ""]
pyr-ASN NAG [spec: 1 "A" 459 ""]
BETA1-4 NAG [spec: 1 "A" 460 ""]
BETA1-4 BMA [spec: 1 "A" 461 ""]
BETA1-6 MAN [spec: 1 "A" 462 ""]
ALPHA1-3 MAN [spec: 1 "A" 463 ""]
ALPHA1-6 FUC [spec: 1 "A" 475 ""]
model file does not exist 
"/software/share/coot-0.9.8.95/coot-gtk2-python/share/coot/data/cho-models/model-level-2-FUC-ALPHA1-6-NAG.tab"
- in create_mmdbmanager_from_res_vector() alt_conf is ""
- in create_mmdbmanager_from_res_vector() use_alt_conf is 0""

--
Markus Meier, Ph.D.
Research Associate
University of Manitoba
Department of Chemistry
144 Dysart Road
Winnipeg, MB, R3T 2N2, Canada
Phone:  +1 204 474 7172
E-mail: markus.me...@umanitoba.ca



To unsubscribe from the COOT list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=COOT=1

This message was issued to members of www.jiscmail.ac.uk/COOT, a mailing list 
hosted by www.jiscmail.ac.uk, terms & conditions are available at 
https://www.jiscmail.ac.uk/policyandsecurity/


[COOT] NTFS compatibility mode on Linux

2024-05-01 Thread Markus Meier

Hi Paul,

The files Coot that creates in coot-backup/ contain colons in their name from 
the timestamp which causes problems on NTFS. Colons are used to select the data 
stream for NTFS files and are therefore disallowed in file names.

Some of our systems are dual-boot for Linux/Windows and our data drives on 
these systems are formatted in NTFS for interoperability. Also our backup 
drives use NTFS format.

Is it possible to turn on NTFS compatibility mode in Coot for Linux to prevent 
the creation of such disallowed file names?

Thanks!
Markus

--
Markus Meier, Ph.D.
Research Associate
University of Manitoba
Department of Chemistry
144 Dysart Road
Winnipeg, MB, R3T 2N2, Canada
Phone:  +1 204 474 7172
E-mail: markus.me...@umanitoba.ca



To unsubscribe from the COOT list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=COOT=1

This message was issued to members of www.jiscmail.ac.uk/COOT, a mailing list 
hosted by www.jiscmail.ac.uk, terms & conditions are available at 
https://www.jiscmail.ac.uk/policyandsecurity/


Re: [COOT] Conflicting opinions on what constitutes an alpha1-6 L-fucose between Coot and Privateer: Building L-fucose fails in coot-0.9.8.94

2024-04-26 Thread Markus Meier

Hi Paul,

I just rebuilt coot-0.9.8.94, coot-0.9.8.93 and coot-0.9.8.92 from scratch, 
deleting the previous installs.

The glyco validator in version coot-0.9.8.94 is indeed fixed. Thank you so much!

Also, coot-0.9.8.93 and coot-0.9.8.92 now build an alpha1-6 fucose correctly 
into the electron density (not sure why I got beta1-6 before), producing 
identical coordinates.

However, building the same fucose in coot-0.9.8.94 fails. First, the carbohydrate moiety 
is inserted into a different orientation compared to coot-0.9.8.93. During the fit, it is 
being moved to the correct density blob, however it is detached completely from the 
carbohydrate tree. Distance between C1 (FUC) and O6 (NAG) results into 3.56 Å! After it 
is built, pressing "Refine tree" does not include the fucose and also exporting 
the tree does not contain it.

I could trace the problem partially down to the library file 
share/coot/data/cho-acedrg/FUC-acedrg.cif

If I remove this file from the installation so that coot-0.9.8.94 cannot load 
it, the fucose builds correctly, similar (but not identical) position to that 
from coot-0.9.8.93.

I can get the same result (correct build) by importing 
share/coot/lib/data/monomers/f/FUC.cif *after* FUC-acedrg.cif had been loaded.
I can re-induce broken behaviour (failed build) by re-importing FUC-acedrg.cif

I believe the two libraries overwrite each other. Curiously enough, 
coot-0.9.8.93 contains the same libraries (identical versions), but I don't 
experience this behaviour, i.e. the fucose is built correctly no matter which 
library was loaded last.

I have attached the carbohydrate trees, one built by coot-0.9.8.93, the other 
by coot-0.9.8.94 (detached fucose included).

Btw. one more funny things I noticed was that the Refmac monomer libraries are 
loaded from $cleaned_dir:
coot-0.9.8.94/coot-Linux-GenuineIntel---gtk2-python/share/coot/lib/data

but the acedrg libraries are loaded from $fat_dir:
coot-0.9.8.94/coot-gtk2-python/share/coot/data

(not that it matters  - except for debugging...)

Thanks again!
With best regards,

Markus


On 2024-04-23 20:23, Engin Özkan wrote:

Caution! This message was sent from outside the University of Manitoba.


I see. Thanks, Paul, for the explanation. Of course you are right, and I
see that too despite having the correct ALPHA1-6:

find_glycosidic_linkage_type() for A 911 NAG,A 913 FUC returns "BETA1-6"

This settles it for me.

(I am not sure how Markus got beta-fucoses built by Coot the in the
first place, though.)

Engin

On 4/23/24 8:03 PM, Paul Emsley wrote:

We are talking about different things, I think.

Coot's glyco builder builds alpha 1-6 linked FUC correctly.

Coot's glyco validator used to (until this evening) identify the alpha
1-6 linked FUC as a beta anomer (it no longer does).

See for example the G8 FUC in 8q5u: Coot 0.9.8.94 now prints:

ASN [spec: 1 "A" 297 ""]
    pyr-ASN NAG [spec: 1 "G" 1 ""]
    BETA1-4 NAG [spec: 1 "G" 2 ""]
    BETA1-4 BMA [spec: 1 "G" 3 ""]
    ALPHA1-3 MAN [spec: 1 "G" 4 ""]
    BETA1-2 NAG [spec: 1 "G" 5 ""]
    BETA1-6 MAN [spec: 1 "G" 6 ""]
    BETA1-2 NAG [spec: 1 "G" 7 ""]
    ALPHA1-6 FUC [spec: 1 "G" 8 ""]

Paul.


On 24/04/2024 01:38, Engin Özkan wrote:

CAUTION: This email originated from outside of the LMB:
.-owner-c...@jiscmail.ac.uk-.
Do not click links or open attachments unless you recognize the
sender and know the content is safe.
If you think this is a phishing email, please forward it to
phish...@mrc-lmb.cam.ac.uk
--

Hi Markus,

I am using Coot that comes with CCP4 8, which has FUC
(alpha-L-Fucose) with an oxygen on C1 in the axial position. Which is
the correct form, I believe.

Using Coot 0.9.8.93, I get my alpha1-6 and alpha1-3 (insect) Fucoses
inserted correctly.  I had trouble understanding your description:
You mean you get Coot to insert a FUC with the glycosidic bond going
equatorial, instead of axial, to Fucose?

This is how it works for me:

https://www.dropbox.com/scl/fi/yjbzstojjivmfpqdlz2os/Screen-Recording-2024-04-23-at-7.32.14-PM.mov?rlkey=zvvk7apd19wk61kt21qyc42l5=0


I'll see what happens when I update to the new Coot version where
Paul has switched the logic.


By the way, privateer tells me I have zero errors and the correct
alpha anomer for Fucose.

Fuc913-a6-GlcNAc911-b-ASN258
WURCS=2.0/2,2,1/[a2122h-1b_1-5_2*NCC/3=O][a1221m-1a_1-5]/1-2/a6-b1
...

PDB Sugar Q   Phi Theta    Detected type
Cnf        Ctx  Ok?
        - --    -- ---
---    --    --- -
test    NAG-A-911   0.570    280.48 5.29 beta-D-aldopyranos

Re: [COOT] Conflicting opinions on what constitutes an alpha1-6 L-fucose between Coot and Privateer

2024-04-24 Thread Markus Meier

Hi Engin,

thank you for all the trouble you went through to make this presentation! The 
glycosidic bond in your presentation is indeed correct (alpha1-6).

Not sure how to interpret the terms axial or equatorial; I followed the 
definition from here:
https://www.masterorganicchemistry.com/2018/02/19/the-big-damn-post-of-sugar-nomenclature

However, judging from the drawings here 
(https://chem.libretexts.org/Bookshelves/Organic_Chemistry/Supplemental_Modules_(Organic_Chemistry)/Alkanes/Properties_of_Alkanes/Cycloalkanes/Rings%3A_cis_trans_and_axial_equatorial_relationships),
 yes, the glycosidic bond in my case is equatorial and privateer reports it as 
beta1-6.

Its not urgent at all, but if you find a moment, could you please send me the terminal output of 
Coot it reports regarding your glycosidic linkages? To do that, start Coot from a terminal and load 
your carbohydrate tree. Centre on one of the atoms in your carbohydrate tree, open the "Add 
N-linked Glycan" panel in the Glyco module and press "Update for Current Residue". 
Coot will print debug info on the terminal.
In the unlike case that terminal output is disabled, you can enable it in 
Edit->Preferences->Other->Console->Display state commands in console?->Yes

Could you also send me the coordinates of your carbohydrate tree? It would like 
to load it in my Coot version(s) and check the terminal output.

Thanks again!
With best regards,
Markus

On 2024-04-23 19:38, Engin Özkan wrote:

Caution! This message was sent from outside the University of Manitoba.


Hi Markus,

I am using Coot that comes with CCP4 8, which has FUC (alpha-L-Fucose)
with an oxygen on C1 in the axial position. Which is the correct form, I
believe.

Using Coot 0.9.8.93, I get my alpha1-6 and alpha1-3 (insect) Fucoses
inserted correctly.  I had trouble understanding your description: You
mean you get Coot to insert a FUC with the glycosidic bond going
equatorial, instead of axial, to Fucose?

This is how it works for me:

https://www.dropbox.com/scl/fi/yjbzstojjivmfpqdlz2os/Screen-Recording-2024-04-23-at-7.32.14-PM.mov?rlkey=zvvk7apd19wk61kt21qyc42l5=0

I'll see what happens when I update to the new Coot version where Paul
has switched the logic.


By the way, privateer tells me I have zero errors and the correct alpha
anomer for Fucose.

Fuc913-a6-GlcNAc911-b-ASN258
WURCS=2.0/2,2,1/[a2122h-1b_1-5_2*NCC/3=O][a1221m-1a_1-5]/1-2/a6-b1
...

PDB Sugar Q   Phi Theta    Detected type
Cnf        Ctx  Ok?
        - --    --    ---
---    --    --- -
test    NAG-A-911   0.570    280.48 5.29  beta-D-aldopyranose
4c1    47.88    (n) yes
test    FUC-A-913   0.535    92.53 177.32    alpha-L-aldopyranose
1c4    70.76    (n) yes
...
Wrong anomer: 0

Best,


Engin


On 4/18/24 4:35 PM, Markus Meier wrote:

Hi Engin,

hmm, it appears I was using the monomer library from
https://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot/dependencies/monomers-2023-01-02-23:57:29.tar.gz
(it gets installed automatically with Coot's build-it script).

However, the files in question (FUC.cif, NAG.cif, BMA.cif and MAN.cif)
are identical to current CCP4 8.0.018 version of the library
(installed by CCP4 Update Manager) as well as version ccp4-8.0.018 and
version ccp4-8.0.016 on GitHub
(https://github.com/MonomerLibrary/monomers)

MD5 checksums on these files are
9cfba4ac24f056d648df8227316f788b  monomers/f/FUC.cif
958ac5b19a92ceec604d233457944b2a  monomers/n/NAG.cif
697dd5242334dd99dad9ce43133c6382  monomers/b/BMA.cif
9d68ff54296ef2f4622005efb4eb080b  monomers/m/MAN.cif

If they are different on your machine, please let me know.

I will switch monomer libraries anyway and do some testing.

I also was switching back and forth between my locally compiled coot
(0.9.8.92) and the the CCP4 distributed version (0.9.8.93 EL). I will
check if there is a difference between the two.

Thank you very much for your suggestion!

With best regards,
Markus

On 2024-04-18 13:48, Engin Özkan wrote:

Caution! This message was sent from outside the University of Manitoba.


Hi Markus,

There were reports to the ccp4bb and phenixbb back in 2011 about
something similar. The monomer library files for FUC and/or the linkage
definitions for refmac/phenix were not right. I think. My recollection
is that this was corrected, and I could not reproduce your observation.

I also had private discussions with Garib Murshudov, who were correcting
the glycan monomer problems back in ~2011.

I could find old links to some of the discussions on bulletin board
archives:

https://phenix-online.org/pipermail/phenixbb/2011-November/017958.html

https://www.mail-archive.com/ccp4bb@jiscmail.ac.uk/msg13968.html

I am curious if something got reverted or if you have some ancient
library definitions somewhere. But I hope this gives some ideas.

Best,

Engin


On 4/17/24 3:07 PM, Markus Meier wrote:

Hi Paul,

thank you for im

Re: [COOT] Conflicting opinions on what constitutes an alpha1-6 L-fucose between Coot and Privateer

2024-04-18 Thread Markus Meier

Hi Engin,

hmm, it appears I was using the monomer library from 
https://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot/dependencies/monomers-2023-01-02-23:57:29.tar.gz
 (it gets installed automatically with Coot's build-it script).

However, the files in question (FUC.cif, NAG.cif, BMA.cif and MAN.cif) are 
identical to current CCP4 8.0.018 version of the library (installed by CCP4 
Update Manager) as well as version ccp4-8.0.018 and version ccp4-8.0.016 on 
GitHub (https://github.com/MonomerLibrary/monomers)

MD5 checksums on these files are
9cfba4ac24f056d648df8227316f788b  monomers/f/FUC.cif
958ac5b19a92ceec604d233457944b2a  monomers/n/NAG.cif
697dd5242334dd99dad9ce43133c6382  monomers/b/BMA.cif
9d68ff54296ef2f4622005efb4eb080b  monomers/m/MAN.cif

If they are different on your machine, please let me know.

I will switch monomer libraries anyway and do some testing.

I also was switching back and forth between my locally compiled coot (0.9.8.92) 
and the the CCP4 distributed version (0.9.8.93 EL). I will check if there is a 
difference between the two.

Thank you very much for your suggestion!

With best regards,
Markus

On 2024-04-18 13:48, Engin Özkan wrote:

Caution! This message was sent from outside the University of Manitoba.


Hi Markus,

There were reports to the ccp4bb and phenixbb back in 2011 about
something similar. The monomer library files for FUC and/or the linkage
definitions for refmac/phenix were not right. I think. My recollection
is that this was corrected, and I could not reproduce your observation.

I also had private discussions with Garib Murshudov, who were correcting
the glycan monomer problems back in ~2011.

I could find old links to some of the discussions on bulletin board
archives:

https://phenix-online.org/pipermail/phenixbb/2011-November/017958.html

https://www.mail-archive.com/ccp4bb@jiscmail.ac.uk/msg13968.html

I am curious if something got reverted or if you have some ancient
library definitions somewhere. But I hope this gives some ideas.

Best,

Engin


On 4/17/24 3:07 PM, Markus Meier wrote:

Hi Paul,

thank you for implementing the Glyco module in Coot!

I used its "Add N-linked Glycan" function to build an N-linked glycan
which contains L-fucose with an alpha1-6 glycosidic linkage on the
first N-acetyl glucosamine of the tree.

However, after exporting the tree, I checked it with privateer and
unfortunately privateer flags the L-fucose as having the wrong
anomeric linkage - beta 1-6 :-(
Coot reports the same as being in the alpha 1-6 anomeric form.

I then manually built the opposite anomeric form which privateer
happily accepts as being alpha1-6. However, Coot reports it as beta 1-6.

In my understanding, the alpha configuration of a glycosidic linkage
should have the oxygen at the anomeric centre and the -CH3 group at
opposite faces of the sugar ring, so I have to agree with privateer on
this one.

Please check you code.


I have attached the two glycan trees and here is the output of
privateer and Coot for both. Coot and privateer versions are given at
the bottom.

Btw. Neither privateer nor refmac5 can read the mmCIF files that Coot
produces with the "Extract this Tree" functions. All the
_atom_site.label_*_id entries are empty!

N133_tree_privateer_alpha_anomer_FUC.cif:

  Privateer:
    Detailed validation data
    

    PDB   Sugar   Q Phi    Theta  Detected type
Cnf    Ctx   Ok?
        -  --  -- ---
---  --  ---  -
    N133  NAG-A-459 0.597   63.27    2.85 beta-D-aldopyranose
4c1  28.09   (n)  yes
    N133  NAG-A-460 0.603  145.39    2.15 beta-D-aldopyranose
4c1  36.34   (n)  yes
    N133  BMA-A-461 0.550  333.80    4.39 beta-D-aldopyranose
4c1  45.25   (n)  yes
    N133  MAN-A-462 0.628    6.26    5.03 alpha-D-aldopyranose
4c1  46.22   (n)  yes
    N133  MAN-A-463 0.628    4.87    5.87 alpha-D-aldopyranose
4c1  50.37   (n)  yes
    N133  FUC-A-464 0.615  162.25  174.51 alpha-L-aldopyranose
1c4  31.24   (n)  yes <-

    Partially occupied monosaccharides, if any, are marked with an
asterisk (*)

    SUMMARY:

   Wrong anomer: 0
   Wrong configuration: 0
   Unphysical puckering amplitude: 0
   In higher-energy conformations: 0

   Privateer has identified 0 issues, with 0 of 6 sugars affected.
    Coot output:
    DEBUG:: find_glycosidic_linkage_type() number of sorted distances: 1
     glyco close: 1.486  A 459  C1   to A 133  ND2
   debug:: find_glycosidic_linkage_type() for A 459 NAG,A 133 ASN
returns "pyr-ASN"
    DEBUG:: find_glycosidic_linkage_type() number of sorted distances: 4
     glyco close: 2.397  A 459  C4   to A 460  C1
     glyco close: 2.339  A 459  O4   to A 460  C2
     glyco close: 2.318  A 459  O4   to A 460  O5
     glyco close: 1.387  A 459  O4   to A 460  C1
   debug:: find_glycosidic_linkage_type() for A 459 NAG,A 46

[COOT] Conflicting opinions on what constitutes an alpha1-6 L-fucose between Coot and Privateer

2024-04-17 Thread Markus Meier
 Partially occupied monosaccharides, if any, are marked with an asterisk (*)

SUMMARY:

   Wrong anomer: 1
   Wrong configuration: 0
   Unphysical puckering amplitude: 0
   In higher-energy conformations: 0

   Privateer has identified 1 issues, with 1 of 6 sugars affected.

  Coot:
DEBUG:: find_glycosidic_linkage_type() number of sorted distances: 1
 glyco close: 1.462  A 459  C1   to A 133  ND2
   debug:: find_glycosidic_linkage_type() for A 459 NAG,A 133 ASN returns 
"pyr-ASN"
DEBUG:: find_glycosidic_linkage_type() number of sorted distances: 4
 glyco close: 2.355  A 459  C4   to A 460  C1
 glyco close: 2.316  A 459  O4   to A 460  O5
 glyco close: 2.313  A 459  O4   to A 460  C2
 glyco close: 1.387  A 459  O4   to A 460  C1
   debug:: find_glycosidic_linkage_type() for A 459 NAG,A 460 NAG returns 
"BETA1-4"
DEBUG:: find_glycosidic_linkage_type() number of sorted distances: 3
 glyco close: 2.37  A 460  O4   to A 461  C2
 glyco close: 2.344  A 460  O4   to A 461  O5
 glyco close: 1.401  A 460  O4   to A 461  C1
   debug:: find_glycosidic_linkage_type() for A 460 NAG,A 461 BMA returns 
"BETA1-4"
DEBUG:: find_glycosidic_linkage_type() number of sorted distances: 1
 glyco close: 1.408  A 461  O6   to A 462  C1
   debug:: find_glycosidic_linkage_type() for A 461 BMA,A 462 MAN returns 
"ALPHA1-6"
DEBUG:: find_glycosidic_linkage_type() number of sorted distances: 3
 glyco close: 2.39  A 461  C3   to A 463  C1
 glyco close: 2.389  A 461  O3   to A 463  C2
 glyco close: 1.398  A 461  O3   to A 463  C1
   debug:: find_glycosidic_linkage_type() for A 461 BMA,A 463 MAN returns 
"ALPHA1-3"
DEBUG:: find_glycosidic_linkage_type() number of sorted distances: 1
 glyco close: 1.414  A 459  O6   to A 464  C1
   debug:: find_glycosidic_linkage_type() for A 459 NAG,A 464 FUC returns 
"ALPHA1-6" <-

Program versions:
Coot 0.9.8.93 EL (ccp4) on Gentoo Linux 64 bit
Privateer version MKIV : 06/02/2 CCP4 8.0.019 on Gentoo Linux 64 bit

With best regards,
Markus

--
Markus Meier, Ph.D.
Research Associate
University of Manitoba
Department of Chemistry
144 Dysart Road
Winnipeg, MB, R3T 2N2, Canada
Phone:  +1 204 474 7172
E-mail: markus.me...@umanitoba.ca



To unsubscribe from the COOT list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=COOT=1

This message was issued to members of www.jiscmail.ac.uk/COOT, a mailing list 
hosted by www.jiscmail.ac.uk, terms & conditions are available at 
https://www.jiscmail.ac.uk/policyandsecurity/

N133_tree_privateer_beta_anomer_FUC.cif
Description: CIF chemical test


N133_tree_privateer_alpha_anomer_FUC.cif
Description: CIF chemical test
N133_tree_privateer_alpha_anomer_FUC.cif:

  Privateer:
Detailed validation data


PDB   Sugar   Q PhiTheta  Detected type Cnf   
 Ctx   Ok?
    -  --  --  ---  ---  -- 
 ---  -
N133  NAG-A-459 0.597   63.272.85  beta-D-aldopyranose  4c1  28.09  
 (n)  yes
N133  NAG-A-460 0.603  145.392.15  beta-D-aldopyranose  4c1  36.34  
 (n)  yes
N133  BMA-A-461 0.550  333.804.39  beta-D-aldopyranose  4c1  45.25  
 (n)  yes
N133  MAN-A-462 0.6286.265.03  alpha-D-aldopyranose 4c1  46.22  
 (n)  yes
N133  MAN-A-463 0.6284.875.87  alpha-D-aldopyranose 4c1  50.37  
 (n)  yes
N133  FUC-A-464 0.615  162.25  174.51  alpha-L-aldopyranose 1c4  31.24  
 (n)  yes <-

Partially occupied monosaccharides, if any, are marked with an asterisk (*)

SUMMARY: 

   Wrong anomer: 0
   Wrong configuration: 0
   Unphysical puckering amplitude: 0
   In higher-energy conformations: 0

   Privateer has identified 0 issues, with 0 of 6 sugars affected.
  
  Coot output:
DEBUG:: find_glycosidic_linkage_type() number of sorted distances: 1
 glyco close: 1.486  A 459  C1   to A 133  ND2 
   debug:: find_glycosidic_linkage_type() for A 459 NAG,A 133 ASN returns 
"pyr-ASN"
DEBUG:: find_glycosidic_linkage_type() number of sorted distances: 4
 glyco close: 2.397  A 459  C4   to A 460  C1  
 glyco close: 2.339  A 459  O4   to A 460  C2  
 glyco close: 2.318  A 459  O4   to A 460  O5  
 glyco close: 1.387  A 459  O4   to A 460  C1  
   debug:: find_glycosidic_linkage_type() for A 459 NAG,A 460 NAG returns 
"BETA1-4"
DEBUG:: find_glycosidic_linkage_type() number of sorted distances: 3
 glyco close: 2.347  A 460  O4   to A 461  C2  
 glyco close: 2.346  A 460  O4   to A 461  O5  
 glyco close: 1.398  A 460  O4   to A 461  C1  
   debug:: find_g