Constrained topology for ancestral reconstruction

[Proginski]

Hi,

I have a gene (CDS) and its orthologs from 5 other species.

Using PRANK, I try to retrieve the ancestral sequence that corresponds to the node represented by a circle( MRCA of Skud, Smik, Spar_NCBI and Scer_NCBI)

So I have the species tree and I assume its topology is correct.
However, since it was built from all the orthologs shared by these species, the length of the branches may not be accurate for the gene I am interested in.

I'd like to build a tree based on the alignment of my gene with its orthologs, but I need to be sure that the topology of the new tree remains the same as the species tree. Because if Sarb becomes a sister group to Scer_NCBI, then I won't be able to reconstruct the desired ancestor (MRCA of Skud, Smik, Spar_NCBI and Scer_NCBI). Possible HGTs are consciously neglected.

I'm performing this ancestral reconstruction on many individual weirdly evolved genes.
My concern is that with the option "--constraint_file", the tree must be unrooted, which I do not want.
Plus when I tried a dummy example :

cat alignment.phylip
 4 48
Seq1       ATGCGTACGT AGCTAGCTAG CTAGCTAGCT AGCTAGCTAG CTAGCTAG
Seq2       ATGCGTACGT AGCTAGCTAG CTAGCTAGCT AAAAAAAAAA CTAGAAAA
Seq3       ATGCGTACGT AGCTAGCTAG CTAGCTAGCT AAAAAAAAAA CTAGAATT
Seq4       ATGCGTACGT AGCTAGCTAG CTAGCTAGCT AGCTAGCTAG CTAGCTAT

cat input_tree_pruned_unrooted.nwk 
(Seq1:1, Seq2:1,(Seq3:1,Seq4:1):1);

phyml -i alignment.phylip -d nt -u input_tree_pruned_unrooted.nwk --constraint_file input_tree_pruned_unrooted.nwk

I got :

cat alignment.phylip_phyml_tree.txt
(Seq3:0.04853078,Seq2:0.00180481,(Seq4:0.00000001,Seq1:0.02168961)0.000000:1.81991069);

So it seems the topology has changed.

Am I missing something, or do you have any tip to achieve this task ? 🤔

Best,

[evolbeginner]

Perhaps you forgot to add --ancestral?