MegaD

MegaD predicts disease status from metagenomic taxonomic profiles using a deep neural network (DNN) to capture complex microbiome patterns.

Key Features:

• Deep Neural Network Implementation: Employs a deep neural network (DNN) architecture tailored for multi-level classification of metagenomic data.
• Input Handling: Accepts taxonomic profiles derived from 16S rRNA sequencing or whole-genome shotgun (WGS) sequencing data.
• Model Building and Classification: Constructs and trains the DNN on input taxonomic profiles and applies the trained model to classify unlabeled metagenomic samples into phenotypic or disease statuses.

Scientific Applications:

• Disease status prediction: Predicts host disease or phenotypic status from microbiome taxonomic profiles to link microbial composition with host phenotypes.
• Microbiome–environment interactions: Supports studies aimed at elucidating roles of microbial communities in biogeochemical processes and their impact on environmental or host phenotypes.

Methodology:

Inputs taxonomic profiles from 16S rRNA or WGS sequencing; trains a deep neural network (DNN) performing multi-level classification on these profiles; applies the trained model to classify unlabeled metagenomic samples.