jonahbrooks / poppr Goto Github PK

It can be a simple R function that calls a C function with an #ifdef for openmp and returns TRUE or FALSE.

Implement a sliding window analysis of the index of association and standardized index of association.

Minimum spanning networks in poppr are currently implemented in bruvo.msn and poppr.msn using the igraph package. I have two requests for these networks:

1. Detect ties in edge weights and add them to the network, creating loops.
2. Allow the user to draw the minimum spanning networks according to the results of mlg.filter where assigned multilocus genotypes are either clustered together in the same node or highlighted in a meaningful fasion.

The most recent version of poppr has changes to how automatic testing by TravisCI is done as well as some minor improvements in the code.

Two issues with single population minimum spanning networks from poppr.msn:

1. showplot argument is not passed on to singlepop_msn internal function.
2. The filtered and clone corrected distance matrix is passed, but the filtered and clone corrected genclone object is not passed.

An #ifdef statement was missing in src/msn.c.

The line

#include <omp.h>

Should be

#ifdef _OPENMP
#include <omp.h>
#endif 

This will allow it to compile on systems that don't yet have openmp support (I'm looking at you, default OSX clang).

In these lines in the function bruvo.msn there are two problems:

• The filtering assumes that the mlg slot is of class MLG. This must be forced. Do it like so before line 482:

pop@mlg <- new("MLG", pop@mlg)

• Line 491 assumes that "contracted" mlgs are the ones to use for the MSN. Change this back to the normal method and set the @visible slot to "contracted" within the if statement:

pop@mlg@visible <- "contracted"

The number of features for mlg.filter are many and clutter the man page for mlg. Moving this function to its own man page would allow for easier navigation.

To avoid breaking backwards compatibility, we could specify a new MLG object that would live in the @mlg slot. The slot could take the value of the mlg object OR a numeric vector. This would need to be designed in such a way that the user could either filter or split MLGs, yet still retain the initial, naive MLG definition. What do you think?

Implement the calculation of boostrapped dendrograms for genlight objects using the function aboot. This could use bitwise.dist by default.

The round-robin allele frequency estimator should calculate allele frequencies for the use of pgen, which is then utilized for psex.

For now, psex can be implemented as 1 - (1 - pgen)^nMLG