Welcome, Guest. Please Login
 YaBB - Yet another Bulletin Board
 
 
  HomeHelpSearchLogin  
 
  HyPhy message board › HYPHY Package › HyPhy bugs › Early end to optimization in version 2.00
(Moderators: Sergei, Simon)
‹ Previous Topic | Next Topic ›
Page Index Toggle Pages: 1
Early end to optimization in version 2.00 (Read 2918 times)
nathanclark111
YaBB Newbies
*
Offline


Feed your monkey!

Posts: 6
Early end to optimization in version 2.00
Nov 4th, 2009 at 10:52am  
I have a script that optimizes just fine in versions 0.99 and 1.00.  However, in version 2.00 the optimization frequently stops prematurely at a very low likelihood value (such as -1e+100). Huh

I am running HyPhy version2.0020090803beta  for MacOS(Universal Binary).  I'm able to use version 1.00 as a work-around, but I thought you'd like to know in the interest of development.

I've attached a .tar.gz file containing the script and data.  Just run the "run_this_script" script and the analysis will start. The models "nullA" and "nullB" are run for pairs of genes. Each report line has the model name, #parameters, log likelihood, and some other parameters. NullA and nullB will not have the same max likelihood, but they will be somewhat close. Notice how some model optimizations stop very early.
I've also included a SAMPLE_output file.
Thanks! Smiley
Back to top
 
 
IP Logged
 
Sergei
YaBB Administrator
*****
Offline


Datamonkeys are forever...

Posts: 1658
UCSD
Gender: male
Re: Early end to optimization in version 2.00
Reply #1 - Nov 4th, 2009 at 10:59am  
Hi Nathan,

Thank you very much for reporting the issue.
I would love to identify the problem, but I don't see an attachment with the files I need to recreate it. Could you confirm that the attachment went through, and barring that - e-mail it to me?

Sergei
Back to top
 
Associate Professor
Division of Infectious Diseases
Division of Biomedical Informatics
School of Medicine
University of California San Diego
WWW WWW  
IP Logged
 
nathanclark111
YaBB Newbies
*
Offline


Feed your monkey!

Posts: 6
Re: Early end to optimization in version 2.00
Reply #2 - Nov 4th, 2009 at 12:13pm  
It didn't like the file suffix. Attaching again.
Back to top
 
Multimedia File Viewing and Clickable Links are available for Registered Members only!!  You need to Login Login (27 KB | )
 
IP Logged
 
Sergei
YaBB Administrator
*****
Offline


Datamonkeys are forever...

Posts: 1658
UCSD
Gender: male
Re: Early end to optimization in version 2.00
Reply #3 - Nov 4th, 2009 at 1:56pm  
Hi Nathan,

I fixed an issue (last week), which would cause the behavior you saw. Could you please retry the script on the newest v2.0 build? I ran your example, and the -1e100 and -1e26 cases both disappeared.

Cheers,
Sergei
Back to top
 
Associate Professor
Division of Infectious Diseases
Division of Biomedical Informatics
School of Medicine
University of California San Diego
WWW WWW  
IP Logged
 
Sergei
YaBB Administrator
*****
Offline


Datamonkeys are forever...

Posts: 1658
UCSD
Gender: male
Re: Early end to optimization in version 2.00
Reply #4 - Nov 5th, 2009 at 2:51am  
The results of running the test file using the latest (20091103) build and default OPTIMIZATION_PRECISION.


octamonkey:HYPHY sergei$ HYPHYMP 8 ~/Desktop/UserData/bug_report_091104/run_this_script.hyphy
Gene List:{"0":"Nup75",
"1":"Nup96",
"2":"Nup98",
"3":"Nup107",
"4":"Nup133",
"5":"Nup153"}
The following comparisons will be made:
1      Nup75      Nup96
2      Nup75      Nup98
3      Nup75      Nup107
4      Nup75      Nup133
5      Nup75      Nup153
6      Nup96      Nup98
7      Nup96      Nup107
8      Nup96      Nup133
9      Nup96      Nup153
10      Nup98      Nup107
11      Nup98      Nup133
12      Nup98      Nup153
13      Nup107      Nup133
14      Nup107      Nup153
15      Nup133      Nup153

Beginning Analysis at 2009/11/5 10:23
COMPARISON_1      Nup75      Nup96
model      np      lnL      yint      slope
nullA      30      -13030.7            0.0701562      NA
nullB      30      -13032      0.0732619      NA
END_1
COMPARISON_2      Nup75      Nup98
model      np      lnL      yint      slope
nullA      30      -13829.6            0.0701745      NA
nullB      30      -13838      0.0746634      NA
END_2
COMPARISON_3      Nup75      Nup107
model      np      lnL      yint      slope
nullA      30      -12505.2            0.0701744      NA
nullB      30      -12514.1      0.0649846      NA
END_3
COMPARISON_4      Nup75      Nup133
model      np      lnL      yint      slope
nullA      30      -15020.4            0.0701829      NA
nullB      30      -15054.8      0.0669783      NA
END_4
COMPARISON_5      Nup75      Nup153
model      np      lnL      yint      slope
nullA      30      -12501.9            0.0701555      NA
nullB      30      -12505.2      0.191955      NA
END_5
COMPARISON_6      Nup96      Nup98
model      np      lnL      yint      slope
nullA      30      -15567.3            0.0732539      NA
nullB      30      -15574.4      0.0746451      NA
END_6
COMPARISON_7      Nup96      Nup107
model      np      lnL      yint      slope
nullA      30      -14242.9            0.0732535      NA
nullB      30      -14250.4      0.0649842      NA
END_7
COMPARISON_8      Nup96      Nup133
model      np      lnL      yint      slope
nullA      30      -16758.1            0.0732629      NA
nullB      30      -16791.2      0.0669765      NA
END_8
COMPARISON_9      Nup96      Nup153
model      np      lnL      yint      slope
nullA      30      -14239.6            0.0732545      NA
nullB      30      -14241.6      0.191826      NA
END_9
COMPARISON_10      Nup98      Nup107
model      np      lnL      yint      slope
nullA      30      -15048.9            0.0746646      NA
nullB      30      -15049.4      0.0649947      NA
END_10
COMPARISON_11      Nup98      Nup133
model      np      lnL      yint      slope
nullA      30      -17564.1            0.0746489      NA
nullB      30      -17590.1      0.0669614      NA
END_11
COMPARISON_12      Nup98      Nup153
model      np      lnL      yint      slope
nullA      30      -15045.6            0.0746386      NA
nullB      30      -15040.5      0.191905      NA
END_12
COMPARISON_13      Nup107      Nup133
model      np      lnL      yint      slope
nullA      30      -16240.2            0.0649746      NA
nullB      30      -16265.7      0.0669625      NA
END_13
COMPARISON_14      Nup107      Nup153
model      np      lnL      yint      slope
nullA      30      -13721.7            0.0649963      NA
nullB      30      -13716.1      0.191893      NA
END_14
COMPARISON_15      Nup133      Nup153
model      np      lnL      yint      slope
nullA      30      -16262.4            0.0669801      NA
nullB      30      -16231.3      0.191857      NA
END_15

Start time: 2009/11/5 10:23
Stop  time: 2009/11/5 10:35
Analysis Complete.
Back to top
 
Associate Professor
Division of Infectious Diseases
Division of Biomedical Informatics
School of Medicine
University of California San Diego
WWW WWW  
IP Logged
 
Page Index Toggle Pages: 1
‹ Previous Topic | Next Topic ›
 
  « Home ‹ Board Top of this page

HyPhy message board » Powered by YaBB 2.5.2!
YaBB Forum Software © 2000-2024. All Rights Reserved.

Valid RSS Valid XHTML Valid CSS Powered by Perl Source Forge