Table 1.

Examples of common tasks and ML methods used in microbiome research.

Task	Predictive goal	Method	Reference
Phenotyping	Sponge bacterial density	Random forests	Moitinho-Silva et al. [42]
Phenotyping	Crop productivity	Random forests	Chang et al. [43]
Phenotyping	Food allergy	Recurrent neural network (LSTM)	Metwally et al. [56]
Phenotyping	Disease (inflammatory bowel disease)	Random forests, lasso, elastic nets	Wirbel et al. [40]
Phenotyping	Disease (e.g., cirrhosis, type 2 diabetes, inflammatory bowel disease)	Convolutional neural networks	Sharma et al. [53], Reiman et al. [54, 55]
Microbial feature classification	Microbiome composition	Autoencoder	García-Jiménez et al. [93]
Microbial feature classification	Metabolic profile	Autoencoder	Le et al. [73]
Interaction analysis	Microbe-metabolite interactions	Embedding	Morton et al. [65]
Interaction analysis	Microbe co-ocurrence patterns	Embedding	Tataru and David [66]
Monitoring composition	Response to diet change	Autoencoder	Reiman and Dai [61]