ENH: expose additional IQ-TREE functions

[rmcar17]

• [x] pairwise ml distances (.mldist file). Return in whatever in-memory data format is easiest - I'll need to convert this to a numpy array on my side of things anyway.
• [x] RapidNJ - (take in a vector of strings for taxa names, as well as pairwise distance matrix (vector of vector of doubles?)) return newick string
• [x] Model Finder - return model string for me to parse? what's easiest?

Anything else?

[thomaskf]

ModelFinder function is now available for PiQTREE. Commit: https://github.com/iqtree/iqtree2/commit/80034a3cbc57a19fbd53b200b0e26e584c76f980

[thomaskf]

The function to build the JC distance matrix is now available in C+++ side. Commit: https://github.com/iqtree/iqtree2/commit/9dce368e9dd3a6a1e550f1d9cc479b1b0fed28ca

[thomaskf]

@rmcar17 @GavinHuttley, Could you clarify if we still need the function to report the ML distance matrix (i.e., the .mldist file) after the phylogenetic inference is completed? During the meeting, I understood that the first objective was changed to report the JC distance matrix directly from an alignment without inferring a tree. I’ve just done this.

[thomaskf]

The function to use Rapid-NJ to build a tree from a distance matrix is now available in C++ side. The last commit: https://github.com/iqtree/iqtree2/commit/cf47d3617480fd5f466ec6ad7aee2c7a4b9aecd6

[thomaskf]

The declarations of these three functions:

// Perform phylogenetic analysis with ModelFinder
// on the input alignment (in string format)
// model_set -- a set of models to consider
// freq_set -- a set of frequency types
// rate_set -- a set of RHAS models
// rand_seed -- random seed, if 0, then will generate a new random seed
// output: modelfinder results in YAML format
string modelfinder(vector<string>& names, vector<string>& seqs, int rand_seed = 0,
                   string model_set = "", string freq_set = "", string rate_set = "");

// Build pairwise JC distance matrix
// output: a phylip-formated distance matrix (in string format)
string build_distmatrix(vector<string>& names, vector<string>& seqs);

// Using Rapid-NJ to build tree from a distance matrix
// input: a phylip-formated distance matrix (in string format)
// output: a newick tree (in string format)
string build_njtree(string dist_matrix);

[thomaskf]

If there is any bug, please kindly let me know. I have not tested the functions extensively. Many thanks @rmcar17 !

[rmcar17]

Thanks @thomaskf! I'll get to porting these over to the python side of things soon.

Re: mldist may as well keep as is for now

[GavinHuttley]

We don't need the ml , just the distance matrix