Questions about estimation of the average standard deviation of split frequencies

[danielyao12]

I'm sorry if I posted my questions in unsuitable place. I tried to use the MrBayes for the BUCKy analysis, and my sequence file was 10,000 bp SNPs. I ran the MrBayes version 3.2.7a on centos7 system. But every diagnfreq displayed "Average standard deviation of split frequencies: NA (no splits above min. frequency)". Originally, I thought that it might be because the number of generation was not long enough. Then, I set the ngen to 21,000,000. It still displayed "Average standard deviation of split frequencies: NA (no splits above min. frequency)". My questions are:

1. Are results from this MrBayes analysis reliable that can be used?
2. If the results are not reliable, how should I do to improve my MrBayes analysis? For example, should I change the lset, nchains, or other parameters in the block?

Here is the block file I used.

#nexus
begin mrbayes;
set autoclose=yes nowarn=yes;
execute 10kb_92_subalignment.nex;
lset nst=6 rates=gamma;
prset brlenspr=Unconstrained:Exp(50.0);
mcmc nruns=2 temp=0.2 ngen=21000000 burninfrac=0.1909
Nchains=4 samplefreq=1000 swapfreq=1000 printfreq=50000
mcmcdiagn=yes diagnfreq=50000 filename=10kb_92;
quit;
end;

This block file was edited basing on the BUCKy tutorial. I also searched similar questions in the Google Group of the BUCKy and other websites in Google, but I didn't find useful information. I appreciate if anyone knows what I should do about my questions. Thank you.

Best regards, Xin

[nylander]

Dear @danielyao12

The NA indicates clearly that there are some errors with the analysis.

I was surprised, however, to see that your command example executed in the first place. Lines 7 and 8 lacks the end-of-line semicolons. But maybe that is a cut-and-paste error on this website?

Unfortunately, assuming the commands indeed was correctly executed, I can't give more advise unless seeing the input file.

Yours

Johan