Branch lengths in output newick tree appear to be double distances from distance matrix

[apetkau]

The branch lengths in the output newick tree when performing e.g., gas mcluster appear to be double the distances in the distance matrix.

Example Input

For example, given the following distance matrix:

dists	sample1	sample2	sample3
sample1	0	1	2
sample2	1	0	3
sample3	2	3	0

From https://github.com/phac-nml/gasclustering/blob/main/tests/data/clusters/expected_tree_hamming.nwk

Example output

The below is an example newick tree:

((sample2:1.000000,sample1:1.000000):0.75,sample3:2.500000);

From https://github.com/phac-nml/gasclustering/blob/main/tests/data/clusters/expected_tree_hamming.nwk

Comparison

The distance from the above distance matrix between sample1 and sample2 is 1.

But, if you add up the branch lengths in the newick file between sample1 and sample2 you get 1 + 1 = 2.

Possible cause

A possible cause is in the buildNewick function:

genomic_address_service/genomic_address_service/classes/multi_level_clustering.py

Lines 52 to 59 in ffca4ed

	def buildNewick(self,node, newick, parentdist, leaf_names):
	if node.is_leaf():
	return "%s:%f%s" % (leaf_names[node.id], parentdist - node.dist, newick)
	else:
	if len(newick) > 0:
	newick = f"):{(parentdist - node.dist) / 2}{newick}"
	else:
	newick = ");"

Internal branch lengths are being divided by 2 here, but not branch lengths leading directly to leaves.

Credit

Credit goes to @kylacochrane, @apetkau, and @jrober84 for finding the issue 😄