Poster Abstracts

Individual factors and conditionally rare taxa shape the skin mycobiome dynamics over seasons
Xinzhao Tong, City University of Hong Kong, Hong Kong, China

We performed a temporal analysis of the skin fungal communities (mycobiomes) of 24 Chinese individuals living in different households to study the skin mycobiomes dynamics over four seasons, and the role of conditionally rare taxa (CRTs) in shaping temporal variability. The results indicated that the skin mycobiomes remained considerably stable across both body sites and time at the genus rank, with the common skin fungus Malassezia detected in all samples. Within the same season, the skin communities clustered significantly by individual and household, and the intra-individual dissimilarity was much less than between cohabitants and different individuals. Within an individual, the community dissimilarity increased with time delay, and the individual-specific CRTs contributed up to 34% of the intra-individual temporal community dissimilarity. Although the skin mycobiomes were personalized and some individual-specific signature taxa were identified, the majority of the individuals could not be correctly matched to their own mycobiomes with time intervals between sampling. However, the prediction accuracy either increased or decreased in half of the individuals after CRTs removal, suggesting that some ecological processes, such as the removal or re-acquisition of taxa over time, might play a functional role in the temporal dynamics and stability of skin mycobiomes.


Metagenome Representation with Scalable Reference Graphs
Andre Kahles, Informatik ETH Zurich, Zurich, Switzerland

Building on techniques from genome assembly and text compression, we use succinct data structures to efficiently represent all sequence information in a k-mer based assembly graph, which not only represents single species and their individual relationships but also captures intra-species variability. A set of more than 50,000 different viral genome sequences is compressed by over 80% when stored in the graph instead as raw sequences and by over 50% when including strain annotations. Importantly, the graph is structured as a self-index that can be used for alignment and annotation of reads arising from metagenome sequencing experiments. Hence, our representation is not only sparse but also efficiently searchable. In addition, the index is dynamic and allows efficient extension to new genome sequences without re-computing the whole index. To keep the graph accessible for fast alignment, we have developed a concept to distribute the index over a set of compute nodes minimizing inter-node communication. The nodes of the graph are colored using compressed binary annotation vectors, encoding information such as species, functional elements or other metadata associated to the underlying sequence. The reference graph leverages information from known genomes as well as from the many previous studies, giving access to rare observations not yet present in reference databases. It is designed to integrate further knowledge over time, e.g., to accumulate information over many studies.


Microbial community structure and inter-phyla interactions exhibit distinct patterns in desert marine mats
Aspassia D. Chatziefthimiou, Renee Richer, Anthony G. Hay, Michael F. Graw, Weill Cornell Medicine Qatar & Richer Environments, Doha, Qatar

The harsh conditions of the desert including wide temperature variations, extreme UV radiation and lack of predictable precipitation patterns drive organisms to the limits of biological existence. Marine tidal zones of deserts may be even more inhospitable as they additionally experience desiccation cycles and hyper-salinity. To understand how abiotic factors direct the active community’s stratification, structure, diversity and function, we collected 6 mm core samples along a 170-m tidal transect at the Inland Sea, Qatar. The transect included 3 distinct biomes: sub- and inter-tidal mats as well as sabkha sand crusts. Abiotic factors were also recorded. 16S rRNA gene deep sequencing was performed on RNA extracted from 1-mm layers of each core. Our analysis demonstrates that active communities as a whole, were distinctly separated based on vertical layer, horizontal biome of origin and temperature. Genera of Bacteroidetes, Chloroflexi and Proteobacteria dominated the subtidal and sabkha biomes while cyanobacterial genera were more abundant in intertidal mats. The intertidal biome exhibited the highest degree of diversity (α, β) and evenness. The first 3 layers of the sub- and inter-tidal mat samples shared indicator species composition, which suggests a metabolic and energetic collaboration towards the harnessing of geochemical conditions in the oxygen and sulfur gradients typical of a mat.


Antimicrobial Resistance in Environmental Water Samples from Bogota, Columbia
Adan Ramirez, Corpogen, Bogota, Columbia

The appearance and dissemination of resistant bacteria in hospitals and the environment, due to excessive antibiotic use, pose an undoubted risk to human health. To assess the extent to which resistance genes have spread in different environments, we surveyed microbial communities and the presence of resistance genes in environmental water samples from Bogota, Colombia. Samples from three locations along the Bogota River with different levels of contamination, and wastewaters from three hospitals in Bogota, were evaluated through 16S rRNA gene sequencing to assess bacterial communities, and metagenomic sequencing to catalog resistance genes. The 16S rRNA gene sequence data revealed low bacterial alpha diversity in hospital and river waters that received high levels of contaminants, and a strong similarity in community composition among the hospital samples. Analysis of the metagenomic sequences showed a greater abundance of resistance genes in hospital waters and the most contaminated point of the river. The observed reduction in bacterial diversity, as well as the abundance and apparent origin of resistance genes in contaminated river water reveal the impact of human activity on water sources and the importance of monitoring microbial communities for assessment of risks to public health and the environment.


Microbiome Analyses of Kindergarten Classrooms using QIIME and Phyloseq
Anders Benteson Nygaard, Oslo and Akershus University College, Oslo, Norway

This study investigated the bacterial composition of floor dust rooms in a recently constructed kindergarten. We hypothesized that the three different room types would exhibit differing bacterial composition due to the different intended uses of each room type, and that a development in the bacterial composition could be observed over time, as the building becomes occupied by its primary occupants, children. Samples were collected five times during the course of one year in one kindergarten. Settled floor dust samples were collected from three separate rooms, one kitchen, one toilet, and one activity room. Total DNA was extracted and the bacterial community diversity of the samples was analyzed using 16S rRNA gene sequencing. Sequence data was processed and analyzed using QIIME and Phyloseq.

Results indicate that the bacterial composition in the three rooms varies, with the bacterial composition of the kitchen showing the highest degree of variation from the toilet and the activity room. The bacterial composition for each room remains relatively stable when during the course of the sampling period. We also find a high prevalence of human-health related bacteria, such as Corynebacterium, Staphylococcus, Propionibacterium, and Streptococcus in most of the rooms.


Composition and dynamics of the air microbiome within a zero carbon building
Marcus H. Y. Leung,1 Xinzhao Tong,1 Jimmy C. K. Tong2 and Patrick K. H. Lee1*
1School of Energy and Environment, City University of Hong Kong, Hong Kong
2Building Sustainability Group, Arup, Hong Kong

Zero carbon buildings (ZCBs) employ unique building designs to minimize energy consumption. However, our understanding of the microbiome in ZCBs remains elusive. Here, the air bacterial community of a ZCB in Hong Kong was characterized by targeting the bacterial 16S rRNA gene. Bacteria associated with the outdoors dominated the ZCB microbial community, with a more modest contribution from taxa commonly associated with skin. Differences in community structure associated with days and seasons were detected. Furthermore, time-decay relationships (based on weighted and unweighted UniFrac distances) differed depending on the season and sampling location. Source-tracking approach including potential sources from adjacent and non-adjacent outdoor air, as well as skin of urban Chinese individuals living in Hong Kong, further supported the importance of the Hong Kong outdoor air microbiome in sourcing the ZCB microbiome regardless of season. Overall, the ZCB microbial assemblage detected and its temporal characteristics were similar to that of conventional built environments based on previous works. Chamber-based works and microbial assemblage investigations of other ZCBs will undoubtedly reveal additional insights related to how the microbiome in ZCBs may be influenced by their unique architectural attributes.


Microbial communities of University campuses in Japan
Kohei Ito, Institute for Advanced Biosciences, Keio University, Tokyo, Japan

Microorganisms live in vairous environments including air, water, soil, and built environments. We spend large amounts of time in built environments, and thus microbes around us in the environments should have strong impact on human health and disease.As far as we know, there are few studies about microbiomes of built environments in Japan. Thus, we collected samples from multi-university campuses in Japan. We set up a sampling design that consists of factors such as campus, building, and surface type to study the effect of each factor on the microbial community structure.
Masafumi Harada, Institute for Advanced Biosciences, Keio University, Tokyo, Japan
Machine learning has been used in various fields such as economics, business administration, natural language processing in recent years. The bioinformatics analysis method has much in common with natural language processing. It is considered that microorganisms in cities correlate with human life and urban environment. Meanwhile, since these metagenomes are huge data  in general, it is difficult for humans to judge relationships. Therefore we examined the use of machine learning for the metagenome in the urban environment.


OPERA-MS: a hybrid approach for metagenomics assembly
Amanda Ng and Eileen Png, Genome Institute of Singapore, Singapore

Motivation: For complex microbial communities, the analysis of information rich, whole-community shotgun sequencing datasets is often restricted by the fragmentary nature of the assembly. The increasing availability of long-reads from third-generation sequencing technologies (e.g.PacBio or OxfordNanopore) can help improve assembly quality, though high error rates and low throughput have limited their application in metagenomics.
Result: In this work, we describe a framework for metagenomics hybrid assembly. This framework combines the advantages of short and long-read technologies, providing high-contiguity and base-pair level accuracy for the reconstructed metagenome. The proposed approach (OPERA-MS) relies on a novel assembly-based metagenome clustering technique and an exact scaffolding-algorithm that can efficiently assemble repeat rich sequences (OPERA-LG[1]).
Based on evaluation on mock communities datasets, we show that the approach used in OPERA-MS leads to significantly more accurate assemblies. OPERA-MS assembles near complete genomes for species with relative abundance >10% and provides high contiguity assemblies (corrected contig N50>500kbp) for species with abundance as low as 1% in complex metagenomics datasets.


Assessing the aerosol microbiome of military relevant environments
James Taylor and Anthony Messer, Defense Science and Technology Laboratory (DSTL), United Kingdom

Dstl has an established aerosol test and evaluation capability; however the understanding of potentially interfering ambient aerosols in evolving militarily-relevant operational environments is still immature. This was acknowledged in the 2012 UK Government Office for Science review of wide area bio-detection. A key recommendation was the need to “develop the ability to sample the air efficiently and assess its diverse composition in representative environments in order to improve the overall sensitivity and selectivity of any future sensing system.”

This poster describes planned aerosol sampling that will be conducted at a greater number of diverse military environments than previous studies. The ambient bioaerosols are to be analysed using high throughput, next generation sequencing technology to give greater confidence in the species level identification of the bacterial content than previous DNA analysis techniques. The measured DNA sequences will be analysed using Bioinformatic tools that have been optimised using in-silico data.
This current study will be used to enhance Dstl’s sovereign capability for aerosol test and evaluation. A standard test mix will be developed to produce a synthetic background aerosol simulating the aerosol characteristics that have been obtained in relevant operational environments.


Automated Purification of Nucleic Acid from Environmental and Clinical Samples for Metagenomic Analysis
Sarah Teter, Promega, USA

In order to gain an accurate representation of the species present in a clinical or environmental sample by next-generation sequencing, it is important to ensure that the nucleic acid extraction is robust across all microbial species and that the resulting purified nucleic acid is of high quality. At Promega, the Scientific Applications group has worked to develop automated methods for purification of nucleic acid from a variety of difficult sample types on the Maxwell® RSC instrument, including fecal specimens, soil samples, and environmental swabs. First, we will describe the results for 16S sequencing of DNA purified from fecal samples using our automated method. The species represented in the 16S sequencing data for the purified DNA showed concordance with published microbial profiles for fecal samples, indicating effective DNA extraction and purification. Next, we will present preliminary results of qPCR assays performed on DNA extracted from soil samples. Finally, we discuss work done with the MetaSUB Consortium to develop an automated method for DNA extraction from environmental swabs collected in transport media. We have recently expanded on this work to identify a method for purification of both DNA and RNA from swabs. These studies present the Maxwell® RSC System as a flexible platform for automated purification of high-quality nucleic acid from a variety of environmental and clinical samples for downstream metagenomic analysis.


Assessment of Urban Microbiome Assemblies with the Help of Targeted Mock Communities
Samuel Gerner, FH Campus Wien University of Applied Sciences

Urban microbiomes are characterized by their comparatively high population dynamics, especially when considering public transport sites such as subway systems with a high fluctuation of bypassing humans. To detect novel species and to enable a detailed analysis of microbe-microbe or host-microbe interactions in such communities, metagenomic reads have to be assembled into ideally complete genomes.

In this study we aim to assemble urban metagenome datasets originating from a pilot study in the subway system of Vienna and from the CAMDA MetaSUB Inter-City Challenge to assess the quality and taxonomic as well as the functional content. To analyse the performance of assembly based methods on urban metagenomic communities, we created mock communities with varying complexity, sequencing depth and quality based on taxonomic profiles of the provided urban metagenome samples. Urban metagenomes were not previously analysed using assembly based methods to our knowledge, thereby mock communities can be used to demonstrate the general applicability of these methods on such communities, enabling thorough testing of the applied assembly methods to samples with known truth. These mock community assemblies are used to propose a set of recommendations for sequencing parameters to obtain optimal assembly and binning quality of urban metagenome data.


Talk Abstracts

Investigating the dynamics of ancestral inference from public surfaces
Eran Elhaik, University of Sheffield, United Kingdom

Microbiome investigations involve scrapping DNA from public surfaces, which oftentimes captures human DNA. Understanding the origins of that DNA is of major interest to the public and researchers. As a topic that most people are familiar with and have some a priori expectation, it is one the most sought for analyses that can be readily interpreted. For researchers, ancestral analysis can shed light on microbial special dynamics and assist in taxonomical annotation. At the moment, our ability to infer ancestry from public surfaces is fraught with challenges, mainly due to the low yield of
human DNA, which reduces the accuracy of the test, and our inability to separate individuals’ DNA from the “communal DNA,” which may not match the individual’s expected ancestry. We overcame the first challenge by developing an extensive gene pool model with millions of 1000 Genomes SNPs. This allows inferring the ancestry relative to the same gene pools, even though samples differ in the number of SNPs captured with one another. To overcome the latter challenge, we developed a simulation of human interactions where the reported ancestry is considered the initial condition before mixing occurs. We attempted to infer the theoretical expected ancestry from this simulation.


The MetaSEW and MetaBEA projects from Montevideo
Gaston Gonnet, Informatik ETH Zurich and Institut Pasteur Montevideo, Zurich, Switzerland and Montevideo, Uruguay

The MetaSEW (metagenomics of the sewage system) and MetaBEA (metagenomics of beaches) are the two main projects of the Centro Uruguayo de Metagenómica. In both cases we have done some preliminary analysis and have promising results.  We have shown the feasibility of several of the goals of both projects.  We think it is very important that we can deliver tangible benefits to the population through these studies.  In turn this means that this generates support from the authorities. We will describe these benefits as they may be useful for other projects.


Challenges associated to the analysis of massive metagenomic data and how could we face them
Nicolas Rascovan, Aix-Marseille University, Marseille, France

Dealing with massive metagenomic datasets, such as those generated by MetaSUB, is extremely challenging at many levels. Some challenges are simply a matter or resources and infrastructure, such as storage, processing power, memory and bandwidth to transfer data. Others are associated to data and associated metadata organization in a uniform, transversal and dynamic way, which is essential for exploring sequencing data. But probably the most significant ones are those associated to the data analysis and how to extract meaningful information out of a vast sea of data. In other words, to identify which are the most interesting scientific questions that could be addressed with MetaSUB data and how should they be tackled from a methodological point of view. Moreover, as a world-wide and massive consortium, we will have to face challenges associated to the distribution and organization of the analysis work, the development of collaboration and cooperation circuits and the coordination of efforts distributed across an extensive transnational scale. This presentation aims at displaying different perspectives and methodological approaches that could be used in the data analysis and to promote open discussions regarding how could we get the best of the sequencing data considering all the challenges mentioned above.


Expanding Multi-omics with NovaSeq™ Systems
Pawel Zajac, Illumina

Next-generation sequencing (NGS) has led to a deeper understanding of the types and functions of microorganisms that are all around us. The rapid technology advances in the last years have enabled large-scale projects and multi-omics studies, where different types of data are combined to uncover features of biological processes that a single data type cannot. In this presentation, I will introduce the multi-omics approach from a metagenomics and metatranscriptomics perspective. In addition, I will show how our flexible and scalable NovaSeq™ sequencing systems fit into this approach, supporting a broad range of applications. Lastly, I will highlight tools and resources to assist you with your next project.