Per taxon sample size

[Matthew-Holding]

Hello,

I notice that the simulations and various empirical examples you have run for TWISST all use multiple samples per designated taxon.

My question is simple: Is there reason to think TWISST would be limited, or at least still useful, when there is only one diploid individual per designated taxon. I have the diploid genomes of four species sequenced, but have only those genomes for variant calling (i.e. no resequenced individuals to increase sample size within species).

Thank you!

[simonhmartin]

Hi, the only real limitation is the amount of information you have.

If you consider the case of just one haploid sequence per taxon, then the weighting for any given topology can only be 1 or 0, so the plot of weightings is really just like painting each chromosome by which tree is observed at each window. If your taxa are closely related, this will be highly dependent on the samples you happened to sequence. Two genomes per taxon gives some quantitative information, but not much, so you might still have several 'peaks' that you don't know whether to trust. As you add genomes, the weightings become more meaningful as they indicate how often each topology is observed at a given part of the genome. So in summary, no real problem with small sample sizes, but you are on the less favourable end of the signal-to-noise ratio. I would say it's still worth a try though.

[Matthew-Holding]

Hi Simon,

Thank you for the thoughtful and rapid reply. It sounds like it is worth a try for me, so I'll give it a go and see what we get.

Cheers,
Matt