Ananke: temporal clustering reveals ecological dynamics of microbial communities

Abstract

Taxonomic markers such as the 16S ribosomal RNA gene are widely used in microbial community analysis. A common first step in marker-gene analysis is grouping genes into clusters to reduce data sets to a more manageable size and potentially mitigate the effects of sequencing error. Instead of clustering based on sequence identity, marker-gene data sets collected over time can be clustered based on temporal correlation to reveal ecologically meaningful associations. We present Ananke, a free and open-source algorithm and software package that complements existing sequence-identity-based clustering approaches by clustering marker-gene data based on time-series profiles and provides interactive visualization of clusters, including highlighting of internal OTU inconsistencies. Ananke is able to cluster distinct temporal patterns from simulations of multiple ecological patterns, such as periodic seasonal dynamics and organism appearances/disappearances. We apply our algorithm to two longitudinal marker gene data sets: faecal communities from the human gut of an individual sampled over one year, and communities from a freshwater lake sampled over eleven years. Within the gut, the segregation of the bacterial community around a food-poisoning event was immediately clear. In the freshwater lake, we found that high sequence identity between marker genes does not guarantee similar temporal dynamics, and Ananke time-series clusters revealed patterns obscured by clustering based on sequence identity or taxonomy. Ananke is free and open-source software available at https://github.com/beiko-lab/ananke.

TaxAss: Leveraging a Custom Freshwater Database Achieves Fine-Scale Taxonomic Resolution

“…The finding that most OTUs match their ecosystem-specific reference sequences with such a high percent identity suggests that the commonly chosen 97% sequence identity clustering is too coarse to observe fine-resolution dynamics. This is supported by previous findings that sequence identity-based OTUs can impose artificial delineations between organisms that affect results differently depending on the lineage ( 33 ) and that sequence identity-based OTUs can contain temporally discordant sequences ( 34 ). We recommend that users planning a taxonomy-centric analysis classify unique sequences after quality trimming and use fine-level taxonomic assignments to group their data instead of sequence identity cutoffs.…”

“…The freshwater tag data sets used in this paper are all publicly available on the National Center for Biotechnology Information’s (NCBI) Sequence Read Archive (SRA). The accession numbers are as follows: Lake Mendota, ERP016591 ( 34 ); Trout Bog, ERP016854 ( 22 ); Lake Michigan, SRP056973 ( 20 ); and Danube River, SRP045083 ( 21 ). The Lake Michigan and bog project accession numbers include additional sample types, so only the Lake Michigan and Trout Bog samples were used.…”

Connections between freshwater carbon and nutrient cycles revealed through reconstructed population genomes

“…The observed taxonomic compositions are consistent with other 16S-based studies from these 225 lakes (Hall et al, 2017;Linz et al, 2017). The detection of similar phyla using both methods 226 suggests that our MAGs are representative of the resident microbial communities.…”

Peer Review #2 of "Freshwater carbon and nutrient cycles revealed through reconstructed population genomes (v0.1)"

“…We also compared 16S rRNA gene amplicon sequencing data from the same timeframe as the metagenomes to confirm that the microbial community composition for these lakes and years was not "abnormal" compared to previous published studies ( Figure S3). The observed taxonomic compositions were consistent with other 16S-based studies carried out on these lakes (Linz et al, 2017;Hall et al, 2017) and with freshwater bacterial community compositions in general (Newton et al, 2011).…”

“…Mendota and Trout Bog Lake are ideal sites for comparative time series metagenomics because of their contrasting limnological attributes and their history of extensive environmental sampling by the North Temperate Lakes -Long Term Ecological Research program (NTL-LTER, http://lter.limnology.wisc.edu) ( Table 1, Table S1). They have also been the subjects of many prior efforts to document and understand freshwater bacterial community diversity and dynamics (Shade et al, 2007;Linz et al, 2017;Hall et al, 2017). We described both predicted pathways in metagenome-assembled genomes (MAGs) and the distributions of functional marker genes to provide a comprehensive overview of microbially-mediated biogeochemical cycling in these two contrasting freshwater lakes.…”