Proportion-based normalizations outperform compositional data transformations in machine learning applications

Microbiome

Table 2 Tree descriptions

	Tree name	Num. nodes	Num. tips	Ave. branch length	Variance branch length	Ultrametric (root-tip distance equal for all tips)
Jones	SILVA	1132	1133	0.0311	0.0012	FALSE
	Filtered_SILVA	75	76	0.0556	0.0032	FALSE
	Filtered_UPGMA	227	228	0.0307	0.0015	TRUE
	UPGMA	28,025	28,026	0.0384	0.0104	TRUE
	IQTREE	28,024	28,026	0.1205	0.1798	FALSE
	Filtered_IQTREE	227	228	0.0840	0.0191	FALSE
Vangay	SILVA	906	907	0.0344	0.0012	FALSE
	Filtered_SILVA	35	36	0.1038	0.0130	FALSE
	Filtered_UPGMA	70	71	0.1300	0.0642	FALSE
	UPGMA	6821	6823	0.0521	0.3193	FALSE
	IQTREE	6821	6823	0.0202	0.0299	FALSE
	Filtered_IQTREE	70	71	0.0813	0.0144	FALSE
Zeller	SILVA	1490	1491	0.0309	0.0009	FALSE
	Filtered_SILVA	121	122	0.0606	0.0044	FALSE
	Filtered_UPGMA	207	208	0.0332	0.0012	TRUE
	UPGMA	11,077	11,078	0.0154	0.0007	TRUE
	IQTREE	11,076	11,078	0.0298	0.0409	FALSE
	Filtered_IQTREE	207	208	0.0792	0.0136	FALSE
Noguera-Julian	SILVA	1233	1234	0.0330	0.0011	FALSE
	Filtered_SILVA	52	53	0.0840	0.0089	FALSE
	Filtered_UPGMA	122	123	0.0273	0.0025	TRUE
	UPGMA	20,365	20,366	0.0509	0.0180	TRUE
	IQTREE	20,364	20,366	0.1204	0.1501	FALSE
	Filtered_IQTREE	122	123	0.0668	0.0295	FALSE

ISSN: 2049-2618