pairwise estimates of dN and dS using runmode = -2 in codeml

[mike....@noaa.gov]

Dear PAML users,

I've occasion to be using PAML again after some time off and seem to be having difficulty in specifying the correct model and Nsites value to do pairwise comparisons among lineages in my tree. I get error messages whenever I specify runmode =-2 with whatever model or Nsites values are in the control file, and the output files 2ML.dS and 2ML.dN aren't generated.

I'd value any suggestions you may have for these parameters or others that will enable me to estimate dN/dS among the lineages and do pairwise estimations between each pair.

thanks!
Mike

[Ziheng]

The pairwise analysis works under model M0 only.
Ziheng

[yasir....@gmail.com]

Dear Prof. Yang,

I am just beginning to use PAML so please excuse me if the question is trivial. Is there a reason why pairwise analysis only works for Nsites=0? I understand that the branch model (model=0) is needed because there are no branches, but different sites could have different substitution models even for pairwise comparisons.

Best,

yasir

[Ziheng]

You can use
runmode = 0
Clock =1
And then specify the NSsites model. The program will allow this, but you will need lots of sites or codons to get reasonable estimates.
Ziheng

[James Jinn-Jy Lin]

Hi Prof. Yang,

I reply on this thread because this thread is relevant to what I wanted to ask. 

I tried the lysozyme example (small dataset) in PAML with runmode =-2. 

And I saw some outputs like the following:

pairwise comparison, codon frequencies: F3x4.

2 (Hla_gibbon) ... 1 (Hsa_Human)

lnL = -564.827965

  0.06505  4.36178  0.36576

t= 0.0651  S=   108.3  N=   281.7  dN/dS=  0.3658  dN = 0.0146  dS = 0.0400

3 (Cgu/Can_colobus) ... 1 (Hsa_Human)

lnL = -639.161206

  0.22638  3.36236  1.46333

t= 0.2264  S=   104.0  N=   286.0  dN/dS=  1.4633  dN = 0.0824  dS = 0.0563

3 (Cgu/Can_colobus) ... 2 (Hla_gibbon)

lnL = -640.460984

  0.22584  3.36585  1.45430

t= 0.2258  S=   103.5  N=   286.5  dN/dS=  1.4543  dN = 0.0821  dS = 0.0564

4 (Pne_langur) ... 1 (Hsa_Human)

lnL = -634.045056

  0.21911  4.56635  1.68484

t= 0.2191  S=   109.3  N=   280.7  dN/dS=  1.6848  dN = 0.0824  dS = 0.0489

I just want to ask whether the dN and dS in the parts labeled by red were calculated using Goldman and Yang (1994).

Thanks!

Jinn-Jy

[Ziheng]

yes, 

runmode =-2

means the ML method from goldman yang 1994.

ziheng