VAMPS: a website for visualization and analysis of microbial population structures

Abstract

Background: The advent of next-generation DNA sequencing platforms has revolutionized molecular microbial ecology by making the detailed analysis of complex communities over time and space a tractable research pursuit for small research groups. However, the ability to generate 10⁵-10⁸ reads with relative ease brings with it many downstream complications. Beyond the computational resources and skills needed to process and analyze data, it is difficult to compare datasets in an intuitive and interactive manner that leads to hypothesis generation and testing.

Results: We developed the free web service VAMPS (Visualization and Analysis of Microbial Population Structures, http://vamps.mbl.edu) to address these challenges and to facilitate research by individuals or collaborating groups working on projects with large-scale sequencing data. Users can upload marker gene sequences and associated metadata; reads are quality filtered and assigned to both taxonomic structures and to taxonomy-independent clusters. A simple point-and-click interface allows users to select for analysis any combination of their own or their collaborators' private data and data from public projects, filter these by their choice of taxonomic and/or abundance criteria, and then explore these data using a wide range of analytic methods and visualizations. Each result is extensively hyperlinked to other analysis and visualization options, promoting data exploration and leading to a greater understanding of data relationships.

Conclusions: VAMPS allows researchers using marker gene sequence data to analyze the diversity of microbial communities and the relationships between communities, to explore these analyses in an intuitive visual context, and to download data, results, and images for publication. VAMPS obviates the need for individual research groups to make the considerable investment in computational infrastructure and bioinformatic support otherwise necessary to process, analyze, and interpret massive amounts of next-generation sequence data. Any web-capable device can be used to upload, process, explore, and extract data and results from VAMPS. VAMPS encourages researchers to share sequence and metadata, and fosters collaboration between researchers of disparate biomes who recognize common patterns in shared data.

Figure 1

The VAMPS website is an interactive data exploration tool that promotes iterative analysis. Users select the datasets and taxonomic levels and classes to analyze, then visualize their microbial community structures using any of a suite of metrics including heatmaps, PCoA plots, bar graphs, pie charts and dendrograms, as well as tables of membership abundance or sequence distributions. Results from initial analyses help refine the further data selection and analytical methods. Users can download graphics, tables, data matrices, tree files, and fasta sequence files for additional analyses and for publication.

Figure 2

The ability to refine the taxonomic selection facilitates exploration of both the more abundant and the less abundant taxa. We illustrate this capability with bar charts, but these selection options are available for other analyses. A demonstrates an initial view of the moderately abundant (> = 1%) bacteria in the public water at a Falmouth, MA distribution point during October and November of 2011. Sphingomonas is the vast majority of all but one of these 5 datasets and is masking the other bacteria. To better identify the patterns within the rest of the microbial community, we can remove the Sphingomonadaceae family (B). To optimize visual interpretation, the selected taxa are drawn to fill the bar graph width, but a mouse-over pop-up displays the true count of reads and relative abundance (of all taxa in the dataset, not of the subset of taxa currently displayed), the full taxonomy, and the dataset name. We have only shown the abundance and genus name here due to space considerations. We can focus in even further by looking only at taxa at less than 10% abundance in any of these datasets, still excluding Sphingomonadaceae (C). Or, we can look at all genera within the order Burkholderiales which appear to be the next most abundant group of taxa after Sphingomonadaceae (D).

Figure 3

To illustrate the use of VAMPS, we step through a simplified analysis process using water samples taken at the North Falmouth Fire Station in Falmouth, MA, project RARE_NFF_Bv6v4. Samples were collected monthly over the course of a year (dataset names have been simplified for display). A bar chart of all taxa at the genus level (A) shows several consistent, abundant taxa and a large number of rare taxa particularly during May, June, August, and September. Holding the cursor over a section reveals the taxon name and abundance. For instance, the light green in the upper left is Flavobacterium, the ochre in the middle is Sphingomonas, and the lavender is Undibacterium. A frequency heatmap (B) of abundant taxa (>3% relative abundance in at least one sample) shows a dominant pattern of Sphingomonas, with a lower abundance in the late summer (July-September) than the rest of the year. Oddly, the months of October and November do not cluster together, nor does October cluster with September, or November with December. A graph of Sphingomonas abundances (C) shows a spike in abundance in October and November, with a gradually decreasing abundance throughout the rest of the year. If Sphingomonas is removed from the analysis and the data are reanalyzed in a new dendrogram, the datasets now cluster by season in three groups: late winter to early spring, late summer to fall, and the two transition times of May-June, and November-December (D). Clades are neatly defined by pairs of subsequent months. This implies a possibility of two different microbial community patterns superimposed on one another that warrants further exploration. Finally, the sampling depth and alpha diversity values are exported to a table for reporting (E, July-December not shown).