Switch to median instead of mean

[sydneyg]

@Microbiology Hi can you please add an option to make a median of the subsampled OTU tables instead of the mean?

[gavinsimpson]

Which functions did you have in mind for this?

[sydneyg]

sorry, I've never written an issue before, this is the avgdist.Rd function.

[gavinsimpson]

This seems trivial to add.

@Microbiology --- as you committed this function, do you have any input on whether this is a useful feature to add (& also #243)? (I'm not expecting you to implement it of course!)

[jarioksa]

The avgdist function does not find the mean of OTU tables, but it calculates the dissimilarities for each subsampled OTU table and returns the average of these dissimilarities. In principle, it could return the median of these dissimilarities, but you were not suggesting this? (And this means having a new trick to calculate such a median).

[jarioksa]

Replacing the line

findist <- Reduce("+", distlist)/length(distlist)

with

findist <- apply(do.call(rbind, distlist), 2, meanfun, ...)

and having new argument meanfun = mean would allow user-selected function to find mean, and having those dots (...) allows mean functions that take extra arguments (such as trim).

[Microbiology]

Thanks @sydneyg for the issue and the feedback (I had suggested this and the other issue be created in a previous conversation).

I will work on adding this in to the avgdist function, using the info from @jarioksa as well. :)

[Microbiology]

I added this functionality and created a pull request to incorporate it (#244).

Also it looks like I am unable to close this issue myself. I think @sydneyg might have to be the one to close it, provided this all looks okay.

[jarioksa]

Closed with PR #244