Hello,

I'd be grateful for some assistance with troubleshooting. I've been running REL though hyphy (HYPHY 2.220140504 beta (MP) for Linux on x86_64) on my desktop ubuntu machine. However, I cannot reproduce the results of running  datamonkey on the same dataset. I expect there would be slight discrepancies, owing to the use of different models, etc, but the results seem to be strongly opposing. That is, on my desktop, I find 5 positively selected codons (BF cutoff = 20), and on datamonkey (Job ID:upload.45868464055980.1), four of these codon sites are found to be negatively selected.

A little more detail on the analyses on my desktop.

I started by estimating marginals. (Note, I made parameter files, and handed them to HYHYMP with system calls in perl, ie HYPHYMP < locus.params > std.out). The contents of the param file is below...

HYPHYMP
2
1
/home/jd/aln/ENSACAP00000021654_exon1_ambig.fa.macse.aln
1
/home/jd/hy/ENSACAP00000021654_exon1_ambig.fa.MG.tree
1
3
2
4
d
3
1
3
3
/home/jd/hy/ENSACAP00000021654_exon1_ambig.fa.rel.out

Then, I used Find Positively Selected Sites, as follows:

HYPHYMP
2
5
7
2
2
2
/home/jd/hy/ENSACAP00000021654_exon1_ambig.fa.rel.out.Dual.marginals
20
/home/jd/hy/ENSACAP00000021654_exon1_ambig.fa.ppr.out

This was the output
Counter Site Index      Log[BF{NS/S>1|s}]
1       84      2.671011887589512
2       167     2.757271429667007
3       185     2.780031678533128
4       257     2.673318104062748
5       283     2.73764052687031


But the datamonkey output looks like this:
Found 5 negatively selected sites (20 significance level)
Codon      E[dS]      E[dN]      Normalized E[dN-dS]      Posterior Probability      Bayes Factor
84      3.261      0.0792843      -3.18172      0.996993      34.7178
167      3.30716      0.0814483      -3.22571      0.997872      49.1188
185      3.27759      0.0796184      -3.19797      0.997315      38.9029
257      3.25518      0.0641413      -3.19104      0.997583      43.2157
283      3.27584      0.0764917      -3.19935      0.997411      40.341

I would be grateful for assistance in finding either 1. the source of my error, or 2. the source of my misunderstanding in interpreting the results.

thanks,
j.