An Integrative Approach for the Functional Analysis of Metagenomic Studies

Abstract

Metagenomics is one of the most prolific "omic" sciences in the context of biological research on environmental microbial communities. The studies related to metagenomics generate high-dimensional, sparse, complex, and biologically rich data-sets. In this research, we propose a framework which integrates omics-knowledge to identify suitable-reduced set of microbiomes features, for gaining insights into functional classification of the metagenomic sequences. The proposed approach has been applied to two Use Case studies, on: 1) cattle rumen microbiota samples, for differentiating nitrate and vegetable oil treated feed, for improving cattle performance, under MetaPlat H-2020 Project 1 , and 2) human gut microbiota and classifying them in functionally annotated categories of leanness, obesity, or overweight. A high Accuracy of 97.5 % and Area Under Curve performance value (AUC) of 0.972 was achieved for classifying Bos taurus, cattle rumen microbiota data samples using Logistic Regression (LR) as classification model as well as feature selector in wrapper based strategy for Use Case 1 and 94.4 % Accuracy with AUC of 1.000, for Use Case 2 on human gut microbiota. In general, LR classifier with Wrapper-LR learner (with ridge estimator) as feature selector, proved to be most robust in analysis.

A metagenomics analysis of rumen microbiome

“…The pipeline was tested on the Bos taurus rumen microbiota samples sequenced for MetaPlat project: -a new data set (SineadNEBQ5) and a data set previously described [15]. The meta-analysis for the categorization of binned metagenomic sequences, known as Operational Taxonomic Units (OTUs) into categories based on feed treated with oil or nitrate in the pipeline was achieved through application of supervised machine learning (ML) technique of classification.…”

“…Optimizing Parameter Settings LR batch size = 100, ridge estimator for log likelihood: 1.0E-8, Conjugate Descent algorithm: True/false, max no of iterations: -1 to 100 NN batch size =100, hidden lay-ers=01/02/no=(attr+classes)/2, value used to seed random number generator (seed)= 1-10, Training Time = 100-500, validation threshold =20, model learning Rate = 0.3, momentum =0. Also, the study for predictive modelling over the use case of Bos taurus was enhanced using feature selection strategies [15]. Wrapper based feature selection strategy with Logistic Regression proved to best in terms of accuracy for analyzing the cattle rumen microbiome at genus level of study [15].…”

“…Also, the study for predictive modelling over the use case of Bos taurus was enhanced using feature selection strategies [15]. Wrapper based feature selection strategy with Logistic Regression proved to best in terms of accuracy for analyzing the cattle rumen microbiome at genus level of study [15]. The improvements achieved with feature selection are indicated in Fig.…”

“…4. The further validations in terms of different sizes of test and train sets for the proposed ML model [15] are indicated in Fig. 5.…”

A Comprehensive Study on Predicting Functional Role of Metagenomes Using Machine Learning Methods

“…Research has also been performed on other varied phenotype hosts such as the ocean, soil, and cattle where microbe-host interactions have great potential in uncovering the influence on an environmental condition being studied [21][22][23][24][25][26][27][28]. Wang et al [26] provided an integrated metagenomic analysis of rumen microbiome responsible for methane emissions and biomass degradation.…”