[R-sig-eco] Phylogenetic turnover using large datasets

[Torsten Hauffe]

Great!

Having a machine with 120 GB of RAM (*envy*), you can certainly use
parallelization to boost the turnover calculation.

Cheers!

On Tue, Mar 10, 2020 at 5:40 PM Jens Ringelberg <jens.ringelberg using gmail.com>
wrote:

> Hi Torsten,
>
> Thank you very much for your answer! You're absolutely right, the problem
> is the number of localities rather than species, so calculating turnover
> separately for each pair of localities is a good (but slow) workaround.
>
> Best,
> Jens
>
> On Fri, 6 Mar 2020 at 09:53, Torsten Hauffe <torsten.hauffe using gmail.com>
> wrote:
>
>> (sorry, I forgot to include the list)
>>
>> Hi,
>>
>> My guess is that the problem is not the size of the phylogeny but the
>> dimension of the community matrix. How many localities and species per
>> locality do you have?
>>
>> Does it work to calculate the dissimilarity between two localities? (Try
>> with the most species-rich)
>> If so, you could either do all pairwise comparisons manually or you
>> replace the memory-hungry apply functions within phylo.beta.pair by
>> (slower) for-loops.
>>
>> Cheers!
>>
>> On Thu, Mar 5, 2020 at 10:06 PM Jens Ringelberg <
>> jens.ringelberg using gmail.com> wrote:
>>
>>> Dear list,
>>>
>>> Does anyone happen to have experience with calculating Simpson's
>>> pair-wise
>>> phylogenetic dissimilarity for large datasets?
>>>
>>> I have tried using betapart’s phylo.beta.pair function to calculate
>>> phylogenetic turnover for a globally-distributed clade with about 1500
>>> taxa, but my machine, which has 125 GB RAM, quickly runs out of memory
>>> while doing so.
>>>
>>> Other than increasing RAM, does anyone have any suggestions for
>>> calculating
>>> Simpson’s dissimilarity for large datasets? I'm aware of the
>>> PhyloMeasures
>>> package, which was designed to work with large datasets, but as far as I
>>> can tell it does not allow one to partition the resulting dissimilarity
>>> index into the true turnover (i.e., Simpson's dissimilarity) and richness
>>> components.
>>>
>>> Many thanks in advance.
>>>
>>>
>>> Jens
>>>
>>> PhD student
>>> University of Zurich
>>>
>>>         [[alternative HTML version deleted]]
>>>
>>> _______________________________________________
>>> R-sig-ecology mailing list
>>> R-sig-ecology using r-project.org
>>> https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
>>>
>>

	[[alternative HTML version deleted]]