modeller comparative
Hi, I am using the ChimeraX modeller comparative command in both ChimeraX 1.7 and ChimeraX 1.11.1. The command executes successfully in both versions, but the results are significantly different. The model generated with ChimeraX 1.7 appears correct, whereas the model generated with ChimeraX 1.11.1 appears incorrect. One notable difference is the GA341 "correlation" score: it is close to 1.0 (high confidence) in ChimeraX 1.7, but close to 0.0 (very low confidence) in ChimeraX 1.11.1. The test data is: Sequence: UniProt P69905 (human hemoglobin) Template structure: PDB 2DHB (horse hemoglobin) Alignment: attached file The exact command used was generated using chimerax GUI: modeller comparative aligned_1.fasta:2 numModels 5 fast false multichain true hetPreserve false hydrogens false waterPreserve false Has anyone observed similar differences between ChimeraX versions, or can suggest a way to obtain consistent Modeller results in the latest releases of ChimeraX? Thanks in advance for any help. Roberto
Hi Roberto, Thanks for alerting us to this problem. It will take me awhile to investigate the differences between what 1.7 is doing and what currently happens. I have opened a ticket in our issue-tracking database (#20506 (Modeller results worse)) <https://www.rbvi.ucsf.edu/trac/ChimeraX/ticket/20506> so that you will receive notifications as progress occurs on resolving the problem. --Eric Eric Pettersen UCSF Computer Graphics Lab
On Jun 17, 2026, at 9:02 AM, Roberto Marabini via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hi,
I am using the ChimeraX modeller comparative command in both ChimeraX 1.7 and ChimeraX 1.11.1. The command executes successfully in both versions, but the results are significantly different. The model generated with ChimeraX 1.7 appears correct, whereas the model generated with ChimeraX 1.11.1 appears incorrect.
One notable difference is the GA341 "correlation" score: it is close to 1.0 (high confidence) in ChimeraX 1.7, but close to 0.0 (very low confidence) in ChimeraX 1.11.1.
The test data is:
Sequence: UniProt P69905 (human hemoglobin) Template structure: PDB 2DHB (horse hemoglobin) Alignment: attached file
The exact command used was generated using chimerax GUI:
modeller comparative aligned_1.fasta:2 numModels 5 fast false multichain true hetPreserve false hydrogens false waterPreserve false
Has anyone observed similar differences between ChimeraX versions, or can suggest a way to obtain consistent Modeller results in the latest releases of ChimeraX?
Thanks in advance for any help.
Roberto <aligned_1.fasta>_______________________________________________ ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
Just to alert the list, a fix for this problem has been committed and will be in tomorrow's daily build. It only affects Modeller jobs where some of the template sequences in the alignment start after the target sequence (i.e. there are gaps to start the template sequences). —Eric
On Jun 17, 2026, at 11:33 AM, Eric Pettersen via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hi Roberto, Thanks for alerting us to this problem. It will take me awhile to investigate the differences between what 1.7 is doing and what currently happens. I have opened a ticket in our issue-tracking database (#20506 (Modeller results worse)) <https://www.rbvi.ucsf.edu/trac/ChimeraX/ticket/20506> so that you will receive notifications as progress occurs on resolving the problem.
--Eric
Eric Pettersen UCSF Computer Graphics Lab
On Jun 17, 2026, at 9:02 AM, Roberto Marabini via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hi,
I am using the ChimeraX modeller comparative command in both ChimeraX 1.7 and ChimeraX 1.11.1. The command executes successfully in both versions, but the results are significantly different. The model generated with ChimeraX 1.7 appears correct, whereas the model generated with ChimeraX 1.11.1 appears incorrect.
One notable difference is the GA341 "correlation" score: it is close to 1.0 (high confidence) in ChimeraX 1.7, but close to 0.0 (very low confidence) in ChimeraX 1.11.1.
The test data is:
Sequence: UniProt P69905 (human hemoglobin) Template structure: PDB 2DHB (horse hemoglobin) Alignment: attached file
The exact command used was generated using chimerax GUI:
modeller comparative aligned_1.fasta:2 numModels 5 fast false multichain true hetPreserve false hydrogens false waterPreserve false
Has anyone observed similar differences between ChimeraX versions, or can suggest a way to obtain consistent Modeller results in the latest releases of ChimeraX?
Thanks in advance for any help.
Roberto <aligned_1.fasta>_______________________________________________ ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
_______________________________________________ ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
participants (2)
-
Eric Pettersen -
Roberto Marabini