bug in CORAL (on win 7, 64 bit)

Interactive modelling (MASSHA, SASpy) and global minimization programs (SASREF, BUNCH, CORAL, GLOBSYMM)
Post Reply
Message
Author
Alex
Active member
Posts: 692
Joined: 2007.08.09 21:10
Location: Planet Earth

bug in CORAL (on win 7, 64 bit)

#1 Post by Alex » 2013.11.12 15:59

Hi
i noticed some strange behavior during CORAL modeling (coral < ans0.txt). Namely, if you check the resulting
model there is a "blank" PDB entry. The coordinates have zero values (compare comp0.pdb and
tmp.pdb).

This is not happening on Mac.

It is also peculiar that CRYSOL actually reads blanks as carbon atoms (tmp00.log vs comp000.log). Is it assuming them to be
ALA (given even that the blank ATOM fields have no ATOM ID, etc)?

Alex
Attachments
coral-bug.zip
(229.26 KiB) Downloaded 297 times

SaxsMax
Administrator
Posts: 349
Joined: 2007.08.03 19:16

Re: bug in CORAL (on win 7, 64 bit)

#2 Post by SaxsMax » 2013.11.15 13:44

Which version of Coral do you use?

I've tried to repeat your run with the same random seed (and it gives the same initial Chi value)
but then SA procedure went different path:
=== CORAL05 started at 15-Nov-2013 12:31:18
Computation mode ....................................... : User
Project identificator .................................. : test
Project decsription .................................... : Will it produce blanc pdb ?
Random sequence initialized from ....................... : 170537
File name with objects info ............................ : t.con
Coordinates of the 1-st subunit evaluated from ......... : AFBEr-Fmod.pdb
15 atoms read, center at 16.58 5.06 -4.67
Subunit was fixed
Coordinates of the 2-nd subunit evaluated from ......... : AFBEr-A.pdb
1897 atoms read, center at 19.04 1.85 0.24
Subunit was fixed
Number of backbone atoms generated ..................... : 259
Averaged formfactors of DRs used
DR formfactor multiplier ............................... : 1.000
Point symmetry of the particle ......................... : P2
Number of equivalent positions ......................... : 2
Cross penalty .......................................... : 7.722e-2
Cross penalty weight ................................... : 100.0
Shift penalty .......................................... : 1.515e-2
Shift penalty weight ................................... : 1.000
Total number of scattering curves ...................... : 1
1-st construct, the first and the last residues: 1 and 259
File name, 1-st experimental data ...................... : NSgpep-extrapC.dat
Number of experimental points found .................... : 1502
Angular units multiplied by ............................ : 0.1000
Number of points reduced to ............................ : 893
Experimental radius of gyration ........................ : 34.52
Number of points in the Guinier Plot ................... : 85
Maximum s-vector in master grid ........................ : 0.2398
Number of points in partial ampliudes .................. : 101
Maximum order of harmonics ............................. : 14
Total penalty .......................................... : 7.737
1-st curve:
NEXP reduced to ........................................ : 892
Theoretical points from 3 to 101 used
The best chi values: 7.25598
Initial fVal ........................................... : 60.39
Spatial step in angstroems ............................. : 5.000
Angular step in degrees ................................ : 20.00
Initial annealing temperature .......................... : 10.00
Annealing schedule factor .............................. : 0.9000
Max # of iterations at each T .......................... : 10000
Max # of successes at each T ........................... : 1000
Min # of successes to continue ......................... : 100
Max # of annealing steps ............................... : 100
==== Simulated annealing procedure started ====
j: 1 T: 0.100E+02 Suc: 1000 Eva: 1418 CPU: 0.299E+02 F:37.4238 Pen: 2.5370
The best chi values: 5.90650
j: 2 T: 0.900E+01 Suc: 1000 Eva: 2867 CPU: 0.605E+02 F:37.4238 Pen: 2.5370
The best chi values: 5.90650
j: 3 T: 0.810E+01 Suc: 1000 Eva: 4397 CPU: 0.928E+02 F:36.6926 Pen: 0.8796
The best chi values: 5.98440

Alex
Active member
Posts: 692
Joined: 2007.08.09 21:10
Location: Planet Earth

Re: bug in CORAL (on win 7, 64 bit)

#3 Post by Alex » 2013.11.15 14:52

hi, I use coral from 2.5.1 ATSAS distribution.
unfortunately, it does not support "coral -v" or "coral -h". In fact it results in a
fortran runtime error if i use any option.

SaxsMax
Administrator
Posts: 349
Joined: 2007.08.03 19:16

Re: bug in CORAL (on win 7, 64 bit)

#4 Post by SaxsMax » 2013.11.15 15:56

Alex wrote:hi, I use coral from 2.5.1 ATSAS distribution.
On which platform?

Also, is the behavior reproducible?

Alex
Active member
Posts: 692
Joined: 2007.08.09 21:10
Location: Planet Earth

Re: bug in CORAL (on win 7, 64 bit)

#5 Post by Alex » 2013.11.15 16:20

On which platform?
see the post subject
Also, is the behavior reproducible?
yes, it was in 10 models. If I use the same input/answer files on Mac, there are no blank ATOM fields
in the file anymore.

SaxsMax
Administrator
Posts: 349
Joined: 2007.08.03 19:16

Re: bug in CORAL (on win 7, 64 bit)

#6 Post by SaxsMax » 2013.11.15 19:53

Tried on Windows 7 with atsas-2.5.1.
And could not reproduce the reported behavior

Alex
Active member
Posts: 692
Joined: 2007.08.09 21:10
Location: Planet Earth

Re: bug in CORAL (on win 7, 64 bit)

#7 Post by Alex » 2013.11.15 20:06

the program persists on my PC.
I attached further OS specs.
Attachments
scr-dump.png
scr-dump.png (50.15 KiB) Viewed 5913 times

SaxsMax
Administrator
Posts: 349
Joined: 2007.08.03 19:16

Re: bug in CORAL (on win 7, 64 bit)

#8 Post by SaxsMax » 2013.11.15 20:15

Please attach your answer file (just in case).

Do you experience similar problem if you run coral with other data (on Win 7) ?

Alex
Active member
Posts: 692
Joined: 2007.08.09 21:10
Location: Planet Earth

Re: bug in CORAL (on win 7, 64 bit)

#9 Post by Alex » 2013.11.15 20:26

Please attach your answer file (just in case).
it is in the *.zip I attached to my first post in this thread.

Do you experience similar problem if you run coral with other data (on Win 7) ?
no, the other jobs are ok.

SaxsMax
Administrator
Posts: 349
Joined: 2007.08.03 19:16

Re: bug in CORAL (on win 7, 64 bit)

#10 Post by SaxsMax » 2013.11.15 21:00

OK, got it.
It happens on Win7 when redirection is used
coral < ans0.txt
Could you please check/confirm, that coral runs properly if the same answers are given in the manual mode?
Thanks

Alex
Active member
Posts: 692
Joined: 2007.08.09 21:10
Location: Planet Earth

Re: bug in CORAL (on win 7, 64 bit)

#11 Post by Alex » 2013.11.18 12:24

Could you please check/confirm, that coral runs properly if the same answers are given in the manual mode?
I confirm that it indeed runs properly when answers are given in manual mode.

Post Reply