PhenoFeatureFinder: a python package for linking developmental phenotypes to omics features

PhenoFeatureFinder is designed to facilitate the analyses required to analyse quantitativeand/or progressive phenotypic- and omics data, and link those using Machine Learning withthe aim to identify causal features, in one package. It can be used for 1) evaluation andvisualisation of phenotype progression over multiple stages and between groups (e.g. treatments,genotypes), 2) pre-processing of omics data, and 3) prediction of features that explain thephenotypic classification. To facilitate usability, each step in the pipeline can also be performedindependently, hence has been assigned a class in the package (Figure 1). We provide anexample of implementation below that focuses on insect development through time and theselection of metabolic features causal to the observed phenotype, but different input datacould be used, provided it has a similar structure. This could be any phenotype that isscored in progressive stages over time. Also, PhenoFeatureFinder was developed initially withmetabolomics data, but users can evaluate its fit applying other types of omics data.