EXPERT: transfer learning-enabled context-aware microbial community classification

Abstract

Microbial community classification enables identification of putative type and source of the microbial community, thus facilitating a better understanding of how the taxonomic and functional structure were developed and maintained. However, previous classification models required a trade-off between speed and accuracy, and faced difficulties to be customized for a variety of contexts, especially less studied contexts. Here, we introduced EXPERT based on transfer learning that enabled the classification model to be adaptable in multiple contexts, with both high efficiency and accuracy. More importantly, we demonstrated that transfer learning can facilitate microbial community classification in diverse contexts, such as classification of microbial communities for multiple diseases with limited number of samples, as well as prediction of the changes in gut microbiome across successive stages of colorectal cancer. Broadly, EXPERT enables accurate and context-aware customized microbial community classification, and potentiates novel microbial knowledge discovery.

Establishment and evaluation of on-chip intestinal barrier biosystems based on microfluidic techniques

“…In addition, the intersection of multiple disciplinary technologies can promote the comprehensive analysis of the on-chip gut biosystems. It is worth noting that artificial intelligence (AI) has great application potential in efficient multi-omics data analysis and microbiome big data mining through digital networks [ 203 ]. On the one hand, AI can also be used to evaluate the success of preset models through computer simulations and provide reliable genotypic and phenotypic data for specific diseases [ 204 ].…”

Emerging applications of artificial intelligence in pathogen genomics

“…Shenhav et al (2019) developed an unsupervised learning approach based on expectation maximisation, which they used to predict the contribution of maternal microflora to infant microbiome, to identify evidence of food and soil contaminants in longitudinal samples from a household, and to distinguish gut microbiota of critically ill patients from those of healthy adults. Deep learning approaches have been successfully applied to classify human microbiomes by the associated disease group with high accuracy and reduced computational requirements for prediction compared to existing approaches (Chong et al, 2022).…”

Deep Learning for Predicting 16S rRNA Gene Copy Number

“…Various deep learning architectures like Convolutional Neural Network (CNN), and Recurrent Neural Network (RNN), have been used for metagenomic-based taxonomy classification (Fioravanti et al, 2018), host phenotype predication (Lo and Marculescu, 2019), disease prediction (Wang et al, 2021), and microbial community predication (Zha et al, 2022). Some other works focus on data mining for antibiotic resistance genes (Arango-Argoty et al, 2018), antimicrobial peptides (Ma et al, 2022), microbial source (Chong et al, 2022), and raw sequencing data (Zhao et al, 2021). All these applications show great potential of deep Learning for microbial studies.…”