indep_c

indep_c calculates the independent contrasts of pqueries.

usage: indep_c [options] placefile[s]

Options

--pp	Use posterior probability for the weight.
-o	Specify the filename to write to.
--out-dir	Specify the directory to write files to.
--prefix	Specify a string to be prepended to filenames.
--csv	Output the results as csv instead of a padded matrix.
--leaf-values	Name of CSV file giving values for the leaves of the tree.

Details

This is an implementation of the “independent contrasts” method of (Felsenstein, 1985).

Assume we have the following pair of placements on a simple tree:

{"tree": "((A:2{0},B:9{1}):7{2},C:5{3},D:1{4}):0{5};", "placements":
  [{"p": [[0, -1309.83, 1, 1, 10]], "n": ["read1"]},
    {"p": [[1, -1309.83, 1, 2, 10]], "n": ["read2"]}
  ], "version": 3, "metadata": {}, "fields":
  ["edge_num", "likelihood", "like_weight_ratio", "distal_length",
    "pendant_length"
  ]
}

And that ic.csv is a (possibly partial) list of tip values as follows:

A,3
B,0
D,1

Then, a run of indep_c will give you the maximum likelihood values of values at the tips assuming that the value changes by Brownian motion:

rhino2 ~/ic » guppy indep_c --leaf-values ic.csv ex.jplace
sequence         x
   read1   2.60377
   read2  0.490566

Note that mft has the --leaf-values option to divide the placement weights by these inferred values.