Hi guys, I was trying out the new Parallel Phaser (dev-1214) and using a model prepared by Rosetta covering 25% of the total structure (32% identity to the homology model), NCS = 2. 5 models actually but not really relevant since the program spontaneously stopped running after a couple of hours (I love what it's supposed to do though!). My question is about the log however. In the Parallel Phaser log it says: Expected LLG and Resolution Limits ---------------------------------- Expected LLG target: 120.00 Expected LLG achieved: 120.0 Resolution of All Data: 2.38 Resolution limited by eLLG target: 4.83 This resolution WILL NOT be used as the default The expected LLG achieved is greater than 120 (eLLG target) but when I tried to recreate that run in the regular Phaser-MR I get this message in the log Expected LLG and Resolution Limits ---------------------------------- Expected LLG target: 120.00 Expected LLG achieved: 49.7 Resolution of All Data: 2.38 Resolution limited by eLLG target: 2.38 This resolution WILL NOT be used as the default The expected LLG achieved is less than 60 (half eLLG target) FTF rescoring WILL be performed due to low expected LLG signal Deep search will be performed from the start I think the second result looks more realistic (unfortunately) but I still don't understand why the difference? Cheers, Morten PS Seems like there's a memory leak in 1214 and 1207 when I run the Phaser step of MR_Rosetta. Does this sound right? Definitely in 1207 but it seems like it's still there in 1214. -- Morten K Grøftehauge, PhD Pohl Group Durham University
Hi, Could you double-check the two log files to see whether there was an input error? I agree that the second eLLG result looks more realistic, so I'm wondering if the same information was given about sequence identity and composition of the asymmetric unit. If all the input is the same, that would mean we have a bug to track down. On the memory leak question, we'll have to look into which version of gcc is used to compile Phaser with OpenMP support. Unfortunately, there do seem to be problems with older versions of gcc (e.g. the 4.2.1 version distributed with Macs) giving memory leaks. The same code compiled with newer versions of gcc (something like 4.4 or newer, if I remember correctly) works fine. Best wishes, Randy On 19 Nov 2012, at 09:45, Morten Groftehauge wrote:
Hi guys,
I was trying out the new Parallel Phaser (dev-1214) and using a model prepared by Rosetta covering 25% of the total structure (32% identity to the homology model), NCS = 2. 5 models actually but not really relevant since the program spontaneously stopped running after a couple of hours (I love what it's supposed to do though!). My question is about the log however. In the Parallel Phaser log it says: Expected LLG and Resolution Limits ---------------------------------- Expected LLG target: 120.00 Expected LLG achieved: 120.0 Resolution of All Data: 2.38 Resolution limited by eLLG target: 4.83 This resolution WILL NOT be used as the default
The expected LLG achieved is greater than 120 (eLLG target)
but when I tried to recreate that run in the regular Phaser-MR I get this message in the log Expected LLG and Resolution Limits ---------------------------------- Expected LLG target: 120.00 Expected LLG achieved: 49.7 Resolution of All Data: 2.38 Resolution limited by eLLG target: 2.38 This resolution WILL NOT be used as the default
The expected LLG achieved is less than 60 (half eLLG target) FTF rescoring WILL be performed due to low expected LLG signal Deep search will be performed from the start
I think the second result looks more realistic (unfortunately) but I still don't understand why the difference?
Cheers, Morten
PS Seems like there's a memory leak in 1214 and 1207 when I run the Phaser step of MR_Rosetta. Does this sound right? Definitely in 1207 but it seems like it's still there in 1214.
-- Morten K Grøftehauge, PhD Pohl Group Durham University
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
------ Randy J. Read Department of Haematology, University of Cambridge Cambridge Institute for Medical Research Tel: + 44 1223 336500 Wellcome Trust/MRC Building Fax: + 44 1223 336827 Hills Road E-mail: [email protected] Cambridge CB2 0XY, U.K. www-structmed.cimr.cam.ac.uk
Randy I'm using gcc 4.6.0 on Mac (from the HPC website http://hpc.sourceforge.net/) Phil On 19 Nov 2012, at 10:28, Randy Read wrote:
Hi,
Could you double-check the two log files to see whether there was an input error? I agree that the second eLLG result looks more realistic, so I'm wondering if the same information was given about sequence identity and composition of the asymmetric unit. If all the input is the same, that would mean we have a bug to track down.
On the memory leak question, we'll have to look into which version of gcc is used to compile Phaser with OpenMP support. Unfortunately, there do seem to be problems with older versions of gcc (e.g. the 4.2.1 version distributed with Macs) giving memory leaks. The same code compiled with newer versions of gcc (something like 4.4 or newer, if I remember correctly) works fine.
Best wishes,
Randy
On 19 Nov 2012, at 09:45, Morten Groftehauge wrote:
Hi guys,
I was trying out the new Parallel Phaser (dev-1214) and using a model prepared by Rosetta covering 25% of the total structure (32% identity to the homology model), NCS = 2. 5 models actually but not really relevant since the program spontaneously stopped running after a couple of hours (I love what it's supposed to do though!). My question is about the log however. In the Parallel Phaser log it says: Expected LLG and Resolution Limits ---------------------------------- Expected LLG target: 120.00 Expected LLG achieved: 120.0 Resolution of All Data: 2.38 Resolution limited by eLLG target: 4.83 This resolution WILL NOT be used as the default
The expected LLG achieved is greater than 120 (eLLG target)
but when I tried to recreate that run in the regular Phaser-MR I get this message in the log Expected LLG and Resolution Limits ---------------------------------- Expected LLG target: 120.00 Expected LLG achieved: 49.7 Resolution of All Data: 2.38 Resolution limited by eLLG target: 2.38 This resolution WILL NOT be used as the default
The expected LLG achieved is less than 60 (half eLLG target) FTF rescoring WILL be performed due to low expected LLG signal Deep search will be performed from the start
I think the second result looks more realistic (unfortunately) but I still don't understand why the difference?
Cheers, Morten
PS Seems like there's a memory leak in 1214 and 1207 when I run the Phaser step of MR_Rosetta. Does this sound right? Definitely in 1207 but it seems like it's still there in 1214.
-- Morten K Grøftehauge, PhD Pohl Group Durham University
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
------ Randy J. Read Department of Haematology, University of Cambridge Cambridge Institute for Medical Research Tel: + 44 1223 336500 Wellcome Trust/MRC Building Fax: + 44 1223 336827 Hills Road E-mail: [email protected] Cambridge CB2 0XY, U.K. www-structmed.cimr.cam.ac.uk
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
Hi Randy,
I'm on
gcc --version
gcc (Ubuntu 4.4.3-4ubuntu5.1) 4.4.3
so that may or may not be the problem.
As far as I can tell the log files are pretty much the same except for the
Wilson scale factor for some reason (but the Wilson scaling seem to be the
same) and then
Select peaks over xx.x% of top (i.e. 0.xxx*(top-mean)+mean)
is different in the two runs (54% and 40.5% respectively).
Cheers,
Morten
On 19 November 2012 11:27, Phil Evans
Randy
I'm using gcc 4.6.0 on Mac (from the HPC website http://hpc.sourceforge.net/)
Phil
On 19 Nov 2012, at 10:28, Randy Read wrote:
Hi,
Could you double-check the two log files to see whether there was an input error? I agree that the second eLLG result looks more realistic, so I'm wondering if the same information was given about sequence identity and composition of the asymmetric unit. If all the input is the same, that would mean we have a bug to track down.
On the memory leak question, we'll have to look into which version of gcc is used to compile Phaser with OpenMP support. Unfortunately, there do seem to be problems with older versions of gcc (e.g. the 4.2.1 version distributed with Macs) giving memory leaks. The same code compiled with newer versions of gcc (something like 4.4 or newer, if I remember correctly) works fine.
Best wishes,
Randy
On 19 Nov 2012, at 09:45, Morten Groftehauge wrote:
Hi guys,
I was trying out the new Parallel Phaser (dev-1214) and using a model prepared by Rosetta covering 25% of the total structure (32% identity to the homology model), NCS = 2. 5 models actually but not really relevant since the program spontaneously stopped running after a couple of hours (I love what it's supposed to do though!). My question is about the log however. In the Parallel Phaser log it says: Expected LLG and Resolution Limits ---------------------------------- Expected LLG target: 120.00 Expected LLG achieved: 120.0 Resolution of All Data: 2.38 Resolution limited by eLLG target: 4.83 This resolution WILL NOT be used as the default
The expected LLG achieved is greater than 120 (eLLG target)
but when I tried to recreate that run in the regular Phaser-MR I get this message in the log Expected LLG and Resolution Limits ---------------------------------- Expected LLG target: 120.00 Expected LLG achieved: 49.7 Resolution of All Data: 2.38 Resolution limited by eLLG target: 2.38 This resolution WILL NOT be used as the default
The expected LLG achieved is less than 60 (half eLLG target) FTF rescoring WILL be performed due to low expected LLG signal Deep search will be performed from the start
I think the second result looks more realistic (unfortunately) but I still don't understand why the difference?
Cheers, Morten
PS Seems like there's a memory leak in 1214 and 1207 when I run the Phaser step of MR_Rosetta. Does this sound right? Definitely in 1207 but it seems like it's still there in 1214.
-- Morten K Grøftehauge, PhD Pohl Group Durham University
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
------ Randy J. Read Department of Haematology, University of Cambridge Cambridge Institute for Medical Research Tel: + 44 1223 336500 Wellcome Trust/MRC Building Fax: + 44 1223 336827 Hills Road E-mail: [email protected] Cambridge CB2 0XY, U.K. www-structmed.cimr.cam.ac.uk
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
-- Morten K Grøftehauge, PhD Pohl Group Durham University
On Mon, Nov 19, 2012 at 1:45 AM, Morten Groftehauge < [email protected]> wrote:
I was trying out the new Parallel Phaser (dev-1214) and using a model prepared by Rosetta covering 25% of the total structure (32% identity to the homology model), NCS = 2. 5 models actually but not really relevant since the program spontaneously stopped running after a couple of hours (I love what it's supposed to do though!). My question is about the log however. In the Parallel Phaser log it says:
By "parallel phaser", do you mean Phaser compiled with OpenMP, or do you mean the GUI that allows you to run multiple searches in parallel? They're very different things... thanks, Nat
GUI alpha-thing.
On 19 November 2012 16:12, Nathaniel Echols
On Mon, Nov 19, 2012 at 1:45 AM, Morten Groftehauge < [email protected]> wrote:
I was trying out the new Parallel Phaser (dev-1214) and using a model prepared by Rosetta covering 25% of the total structure (32% identity to the homology model), NCS = 2. 5 models actually but not really relevant since the program spontaneously stopped running after a couple of hours (I love what it's supposed to do though!). My question is about the log however. In the Parallel Phaser log it says:
By "parallel phaser", do you mean Phaser compiled with OpenMP, or do you mean the GUI that allows you to run multiple searches in parallel? They're very different things...
thanks, Nat
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
-- Morten K Grøftehauge, PhD Pohl Group Durham University
On Mon, Nov 19, 2012 at 10:14 AM, Morten Groftehauge
By "parallel phaser", do you mean Phaser compiled with OpenMP, or do you mean the GUI that allows you to run multiple searches in parallel? They're very different things...
GUI alpha-thing.
Okay, this is different than what Randy or Phil were talking about. I suspect the resolution limits are being handled differently - in the parallel Phaser wrapper, it sets the high resolution for both the MR search and the refinement step to be the same, unlike the standard Phaser-MR GUI where these are clearly separate. I'm not sure if this is optimal behavior - I wrote it this way mostly for speed reasons. At any rate, this program was essentially an experiment that never went anywhere and has effectively been superseded by the newer MR pipeline (phaser.MRage); it might be time to retire it. -Nat
participants (4)
-
Morten Groftehauge -
Nathaniel Echols -
Phil Evans -
Randy Read