Fingerprinting cities: differentiating subway microbiome functionality

Abstract

BackgroundAccumulating evidence suggests that the human microbiome impacts individual and public health. City subway systems are human-dense environments, where passengers often exchange microbes. The MetaSUB project participants collected samples from subway surfaces in different cities and performed metagenomic sequencing. Previous studies focused on taxonomic composition of these microbiomes and no explicit functional analysis had been done till now.ResultsAs a part of the 2018 CAMDA challenge, we functionally profiled the available ~ 400 subway metagenomes and built predictor for city origin. In cross-validation, our model reached 81% accuracy when only the top-ranked city assignment was considered and 95% accuracy if the second city was taken into account as well. Notably, this performance was only achievable if the similarity of distribution of cities in the training and testing sets was similar. To assure that our methods are applicable without such biased assumptions we balanced our training data to account for all represented cities equally well. After balancing, the performance of our method was slightly lower (76/94%, respectively, for one or two top ranked cities), but still consistently high. Here we attained an added benefit of independence of training set city representation. In testing, our unbalanced model thus reached (an over-estimated) performance of 90/97%, while our balanced model was at a more reliable 63/90% accuracy. While, by definition of our model, we were not able to predict the microbiome origins previously unseen, our balanced model correctly judged them to be NOT-from-training-cities over 80% of the time.Our function-based outlook on microbiomes also allowed us to note similarities between both regionally close and far-away cities. Curiously, we identified the depletion in mycobacterial functions as a signature of cities in New Zealand, while photosynthesis related functions fingerprinted New York, Porto and Tokyo.ConclusionsWe demonstrated the power of our high-speed function annotation method, mi-faser, by analysing ~ 400 shotgun metagenomes in 2 days, with the results recapitulating functional signals of different city subway microbiomes. We also showed the importance of balanced data in avoiding over-estimated performance. Our results revealed similarities between both geographically close (Ofa and Ilorin) and distant (Boston and Porto, Lisbon and New York) city subway microbiomes. The photosynthesis related functional signatures of NYC were previously unseen in taxonomy studies, highlighting the strength of functional analysis.

Liquid biopsies and cancer omics

“…The microbiota creates a complex network that can influence the tumor microenvironment in a very heterogeneous way that relies on the intrinsically heterogeneity of the microbiome itself [44][45][46][47][48][49] . The microbiome study, also through liquid biopsies, would also grant some other information related to the geography of diseases, among which cancer, since the intestinal bacteria are able to differentiate individuals on the basis of the place they live in [50][51][52][53][54] .…”

Literature on Applied Machine Learning in Metagenomic Classification: A Scoping Review

“…These conferences provide annual challenges in the field of metagenomics based on the MetaSUB data set of microorganism samples from various subways from all over the world. Works have been done to unravel bacterial fingerprints of city subways from microbiome 16S gene profiles [ 32 , 33 , 34 ] and to show that bacterial composition across different cities is significantly different. This is of crucial importance as it potentially allows us to deduce the location of a given sample, which can have many applications.…”

Metagenomic Geolocation Prediction Using an Adaptive Ensemble Classifier

“…For instance, Casimiro-Soriguer et al ( 2019) used decision trees for the classification of functional profiles created from metagenomics data. The random forest classifier was used by Ryan (2019), Harris et al (2019), and Walker and Datta (2019), while Zhu et al (2019), and Walker and Datta (2019) used the SVM classifier. In some cases, several dimension reduction and feature engineering techniques have been applied in conjunction with the classification algorithms being utilized.…”

“…This was followed by classification to city labels using the random forest. Zhu et al (2019) emphasize the use of functional profiling of microbiome data over taxonomic profiling and applied support vector machines (SVM) to predict geolocation of unknown samples. While Kawulok et al (2019) utilized a k-mer based approach to design fingerprints for the identification of source origins of metagenomics samples.…”