Thom Quinn comments

Results 50 comments of


                                            Thom Quinn

Question about when to use CLR or ILR

Oh, I think `ivar=NA` toggle from issue #9 was never implemented.

Question about when to use CLR or ILR

Ah, I think it is implemented on GitHub only, not CRAN...

Question about when to use CLR or ILR

Thanks for checking. Does this code work? ``` library(propr) data(iris) met.rel

Question about when to use CLR or ILR

Also please let me know what `packageVersion("propr")` says :-)

Question about when to use CLR or ILR

Hm, my guess is that the cluster is not loading 4.2.8 correctly for some reason. In an older version of R, `ivar = NA` would call CLR by default, hence...

Question about scaling propr to very large data sets (Part 2)

Thanks for your interest in propr! It sounds like the data can be thought of as "multi-omics data" in that you have two views of the same subjects. I'll assume...

Question about scaling propr to very large data sets (Part 2)

Ah, I see. Unfortunately, I have not made any functions for this case. The #9 solution would be the way to go. You may find the functions `getMatrix()` and `getRatios()`...

Question about scaling propr to very large data sets (Part 2)

Our reason for calculating FDR on only positive values is that the negative values can be difficult to interpret (see #4). I don't feel comfortable endorsing exact p-value calculation from...

Question about scaling propr to very large data sets (Part 2)

I think the key issue is that the negative rho (like negative correlations) would be highly sensitive to the choice of the reference. They may mean something if you believe...

Question about scaling propr to very large data sets (Part 2)

[s41592-020-01006-1.pdf](https://github.com/tpq/propr/files/6658929/s41592-020-01006-1.pdf)