Welcome, Guest. Please Login
YaBB - Yet another Bulletin Board
 
  HomeHelpSearchLogin  
 
HyPhy vs. PAML (Read 4085 times)
adk
Guest


HyPhy vs. PAML
May 10th, 2005 at 1:17pm
 
Hey Guys,

So here's a question: I've been comparing some output from HyPhy to PAML output, and for some reason things look very different. In particular it seems like branch lengths are very often shorter in HyPhy than in PAML.  To bench mark this I have been using the default (model = 0) settings of GY94 (in HyPhy) to PAML codeml output. I'm using the post_dNdS.bf file to parse branch lengths btw. Any insights?

cheers,
Andy
Back to top
 
 
IP Logged
 
Sergei
YaBB Administrator
*****
Offline


Datamonkeys are forever...

Posts: 1658
UCSD
Gender: male
Re: HyPhy vs. PAML
Reply #1 - May 10th, 2005 at 1:19pm
 
Dear Andy,

Quote:
Hey Guys,

So here's a question: I've been comparing some output from HyPhy to PAML output, and for some reason things look very different. In particular it seems like branch lengths are very often shorter in HyPhy than in PAML.  To bench mark this I have been using the default (model = 0) settings of GY94 (in HyPhy) to PAML codeml output. I'm using the post_dNdS.bf file to parse branch lengths btw. Any insights?



Could you send me the benchmark file and codeml output, so that I can see what's going on? I am guessing this is a scaling issue...

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: HyPhy vs. PAML
Reply #2 - May 10th, 2005 at 2:27pm
 
Dear Andy,

I looked at your example, and it is indeed a scaling issue. PAML output looks like

Note: Branch length is defined as number of nucleotide substitutions per codon (not per neucleotide site).

tree length =   0.50537

(1: 0.097822, 2: 0.095322, 3: 0.312227);

(mel: 0.097822, sim: 0.095322, yak: 0.312227);


HyPhy reports (mel:0.0307338,sim:0.0327225,yak:0.100553).

Effectively HyPhy branch lengths (substitutions per nucleotide site) are 1/3 of what PAML reports (substitutions per codon) site.

Other subtle difference could arise because HyPhy and PAML treat ambiguity data a bit differently as I recall.

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
 
adk
Guest


Re: HyPhy vs. PAML
Reply #3 - May 10th, 2005 at 2:37pm
 
Great Sergei! Where can I find out the details of how HyPhy deals wiwth ambiguous sites?

cheers,
Andy
Back to top
 
 
IP Logged
 
Sergei
YaBB Administrator
*****
Offline


Datamonkeys are forever...

Posts: 1658
UCSD
Gender: male
Re: HyPhy vs. PAML
Reply #4 - May 10th, 2005 at 2:46pm
 
Dear Andy,

Quote:
Great Sergei! Where can I find out the details of how HyPhy deals wiwth ambiguous sites?

cheers,
Andy


Pages 5-6 of Multimedia File Viewing and Clickable Links are available for Registered Members only!!  You need to Login Login. Also, HyPhy includes ambiguities when counting base frequencies (by allocating 1/k count to each possible resolution).

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