ancestral state construction made longer sequences for parents

[abs]

Hello!
I used RAxML -ng to construct ancestral state for a large set of Influenza virus DNA sequences (spike gene). 

my input tree is made by fasttree and the input for MSA is made by mafft and the model for RAxML is GTR+G  . 
but the result is not satisfying! the ancestors that created are larger than the children sequences specially at the first and end of the sequence. while when I align the children with same ancestor they only have few nucleotide difference. I appreciate if you help me to understand how made more accurate tree for constructing ancestors.

[Grimm]

Hi Abbas,

there may be a problem with your MSA. Since the ancestral state reconstruction is site-wise, the reconstructed sequence cannot have a different length unless the sister clades have different lengths.

Note that ML ancestral state reconstruction is not a purely top-down process. So, if the sister clade of X or further away tips have the full spike protein sequence but your children miss the first and last couple one (incomplete sequences), then for the ancestor of X, RAxML gauges the childrens' missing tip state (in a likelihood environment missing and gaps translate into a p(1,1,1,1) tip propability vector) with the information reconstructed for the deeper part of the tree, i.e. ancestral states reconstructed for the sister clade's root etc.

Naturally, such reconstructions involving missing-data tips are tricky. Ideally, you hence should only include sites for the ancestral state reconstruction that are defined for all tips in your data set and make sure that your MSA is properly clipped at the start and end before running the ancestral state reconstruction.

Good reconstructing

/G

[abs]

Hello Grimm,
thanks for the good explanation, it really cleared everything up.

--
You received this message because you are subscribed to a topic in the Google Groups "raxml" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/raxml/cSmDqQdM17Q/unsubscribe.
To unsubscribe from this group and all its topics, send an email to raxml+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/raxml/87dfd200-a706-4396-b498-80c782b61cd6n%40googlegroups.com.

[Xinxin Zhu]

Hello 

Can I seek help from you on the ancestral sequence reconstruction with RaxmL? I failed to find which command line used to perform an ancestral sequence construction in the tutorial. So can you provide help to me? Thank you very much.

[Alexandros Stamatakis]

Are you referring to RAxML or RAxML-NG?

Alexis

> <https://groups.google.com/d/topic/raxml/cSmDqQdM17Q/unsubscribe>.

> To unsubscribe from this group and all its topics, send an email
> to raxml+un...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/raxml/87dfd200-a706-4396-b498-80c782b61cd6n%40googlegroups.com

> <https://groups.google.com/d/msgid/raxml/87dfd200-a706-4396-b498-80c782b61cd6n%40googlegroups.com?utm_medium=email&utm_source=footer>.
>
> --
> You received this message because you are subscribed to the Google
> Groups "raxml" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to raxml+un...@googlegroups.com
> <mailto:raxml+un...@googlegroups.com>.

> To view this discussion on the web visit

> https://groups.google.com/d/msgid/raxml/9b2e6969-afff-4b91-9701-849479ad382dn%40googlegroups.com
> <https://groups.google.com/d/msgid/raxml/9b2e6969-afff-4b91-9701-849479ad382dn%40googlegroups.com?utm_medium=email&utm_source=footer>.

--
Alexandros (Alexis) Stamatakis

Research Group Leader, Heidelberg Institute for Theoretical Studies
Full Professor, Dept. of Informatics, Karlsruhe Institute of Technology
Affiliated Scientist, Evolutionary Genetics and Paleogenomics (EGP) lab,
Institute of Molecular Biology and Biotechnology, Foundation for
Research and Technology Hellas

www.exelixis-lab.org

[HA]

Hello!
I don't know what exact problem you have, but I made it this way.

I had one initial tree , one MSA file then I used them as input files in the following code ( you can use your desirable models in --model part)
code :  <<        raxml-ng --ancestral --msa aligned_HA1 --tree rooted_initial_tree.tree --model GTR+G --prefix ASR   >>

it gives several files and you can see what are these files in a log file that is generated by this command. you can find out more about models that might fit your project in instruction.

best

To view this discussion on the web visit https://groups.google.com/d/msgid/raxml/9b2e6969-afff-4b91-9701-849479ad382dn%40googlegroups.com.