Metagenomic Geolocation Prediction Using an Adaptive Ensemble Classifier

Abstract

Microbiome samples harvested from urban environments can be informative in predicting the geographic location of unknown samples. The idea that different cities may have geographically disparate microbial signatures can be utilized to predict the geographical location based on city-specific microbiome samples. We implemented this idea first; by utilizing standard bioinformatics procedures to pre-process the raw metagenomics samples provided by the CAMDA organizers. We trained several component classifiers and a robust ensemble classifier with data generated from taxonomy-dependent and taxonomy-free approaches. Also, we implemented class weighting and an optimal oversampling technique to overcome the class imbalance in the primary data. In each instance, we observed that the component classifiers performed differently, whereas the ensemble classifier consistently yielded optimal performance. Finally, we predicted the source cities of mystery samples provided by the organizers. Our results highlight the unreliability of restricting the classification of metagenomic samples to source origins to a single classification algorithm. By combining several component classifiers via the ensemble approach, we obtained classification results that were as good as the best-performing component classifier.

Where environmental microbiome meets its host: Subway and passenger microbiome relationships

“…Furthermore, certain variations are intended to reducing the computation time. In particular, instead of using reads directly, researchers used k‐mers, constructed by splitting the reads into smaller sequences such as 24 bp, as classifiers (Anyaso‐Samuel et al, 2021; Huang et al, 2020). Some pipelines reduce the data by excluding features from the abundance tables; for example, some genera and features unable to distinguish samples are excluded (Casimiro‐Soriguer et al, 2019; Walker et al, 2018), or dimension reduction methods, such as principal component analysis (PCA), are used.…”

Prediction of Smoking Habits From Class-Imbalanced Saliva Microbiome Data Using Data Augmentation and Machine Learning

“…Avoiding the pooling of data from different studies can bypass the study-specific effect issue, though greatly reduces the statistical power with negative effects on the reliability of the outcome. Additionally, microbiome data commonly suffer from imbalanced sample distribution (Khan and Kelly, 2020 ; Poore et al, 2020 ; Anyaso-Samuel et al, 2021 ). Particularly in (binary) classification applications, it is commonly the case that one class is overrepresented (majority class) while the other is underrepresented (minority class).…”

Multiblock partial least squares and rank aggregation: Applications to detection of bacteriophages associated with antimicrobial resistance in the presence of potential confounding factors

“…This challenge exhorts a comprehensive examination of anti‐microbial resistance (AMR) patterns in this vast metagenomic surveillance data, attracting dedicated researchers striving to uncover these complex interactions. Unlike previous studies that have predominantly focused on geolocation prediction 4–6 or spatial modelling 7,8 of such patterns, our research forges a new path by delving into the uncharted territory of bacteriophages' role in orchestrating AMR dissemination. Bacteriophages, also known as phages, are viruses that prey on bacteria 9 .…”