Fully integrative species delimitation with machine learning and diverse data types in delimSOM 2.0
Species delimitation increasingly relies on multiple sources of evidence, but most frameworks are still limited to genetics and morphology or analyze data types separately and compare results qualitatively. We present delimSOM 2.0, an R package that leverages unsupervised machine learning through multilayer self-organizing maps. Any kind of input data, including genomic, phenotypic, ecological,…
Species delimitation is becoming more complex as it now relies on multiple sources of evidence. However, most existing frameworks still only use genetics and morphology, or analyze different data types separately and compare the results qualitatively.
The researchers have developed a new R package called delimSOM 2.0. This package utilizes unsupervised machine learning through multilayer self-organizing maps to analyze any type of input data, including genomic, phenotypic, ecological, and spatial data. The package treats each data type as a separate layer and balances their contributions through layer-specific weighting.
DelimSOM 2.0 offers a range of functions to help users preprocess their data, perform replicate training, conduct multivariate clustering, optimize hyperparameters, assess model quality, and analyze variable and layer importance.
To test the effectiveness of this new approach, the researchers used two simulation sets and eight empirical studies covering a wide range of taxa and data types. The results showed that the method performed best when dealing with larger sample sizes (more than 10 individuals per lineage) and when the number of lineages (K) was modest.
Written by urgent.news from bioRxiv's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
