SpeciesRax

[Clifton Lewis]

Hi Benoit,

I had a question regarding SpeciesRax. I was wondering do all the genetrees need to have all the species sampled. I presumed not but for some reason the inferred species tree has for some reason just collapsed most of the nodes. I was wondering what the logic behind that was? I have run with a DTL model as recommended with a starting gene trees provided. Appreciate your help in understanding this.

Thanks,

Clifton

[Benoit Morel]

Hi Clifton,

The question you are asking is a topic I am currently exploring :-)

Missing genes (whose absence is not due to gene loss) might indeed be detrimental to species tree inference. The --pruned-species-tree option is supposed to partly correct this problem, but this solution is not perfect. 

What do you mean with collapsing most of the nodes? Could you maybe send me your result species trees?

Here are some ideas:

- try filtering out gene trees that cover too little species and see if it improves the results

- try competing methods, such as Astral-Pro, to see if they are more robust to missing data (I would expect quartets to behave slightly better

if missing data is the problem).

Best,

Benoit

[Clifton Lewis]

Hi Benoit,

After looking at the input trees I realised I had a mistake in that there were a few underscore in the speices name for a few species that I hadnt corrected for which explains the error. Sorry for the bother.

Thanks,

Clifton