Second Genome Announces Expansion of Scientific Advisory Board

Second Genome today announced the expansion of our Scientific Advisory Board. Our preeminent board of leaders in the genomic technologies and microbiome space now includes: Dr. Gary Andersen, Dr. Martin Blaser, Dr. Michael Fischbach, Dr. Susan Lynch, Dr. Pankaj Jay Pasricha, Dr. David Relman, and Dr. Justin Sonnenburg. Read more about our advisors.

PMA-PhyloChip

PMA-PhyloChip DNA Microarray To Elucidate Viable Microbial Community Structure

On December 1st, 2011 NASA’s Jet Propulsion Laboratory in Pasadena California posted this brief on a technique for differentiating live versus dead bacteria in complex samples using a combination of sample treatment with propidium monoazide (PMA) followed by community characterization with Second Genome's PhyloChip assay. The article is reprinted here with permission from JPL-NASA and Tech Briefs magazine.
You can visit the original article at:
http://www.techbriefs.com/component/content/article/12241

Please contact us if you plan to use or test this approach in your research. There may be special boxes to mark on your Specimen Shipment Forms (SSFs) and Data Analysis Plans (DAPs) to inform our lab of the additive. Currently, Second Genome does not perform the PMA step but we do provide the downstream lab operations and analysis from your isolated gDNA.

Tech Briefs Article:

.

This technology has applications in pharmaceutical and medical equipment manufacturing, and food processing.

Since the Viking missions in the mid- 1970s, traditional culture-based methods have been used for microbial enumeration by various NASA programs. Viable microbes are of particular concern for spacecraft cleanliness, for forward contamination of extraterrestrial bodies (proliferation of microbes), and for crew health/safety (viable pathogenic microbes). However, a “true” estimation of viable microbial population and differentiation from their dead cells using the most sensitive molecular methods is a challenge, because of the stability of DNA from dead cells.

The goal of this research is to evaluate a rapid and sensitive microbial detection concept that will selectively estimate viable microbes. Nucleic acid amplification approaches such as the polymerase chain reaction (PCR) have shown promise for reducing time to detection for a wide range of applications. The proposed method is based on the use of a fluorescent DNA intercalating agent, propidium monoazide (PMA), which can only penetrate the membrane of dead cells. The PMA-quenched reaction mixtures can be screened, where only the DNA from live cells will be available for subsequent PCR reaction and microarray detection, and be identified as part of the viable microbial community. An additional advantage of the proposed rapid method is that it will detect viable microbes and differentiate from dead cells in only a few hours, as opposed to less comprehensive culture-based assays, which take days to complete. This novel combination approach is called the PMA-Microarray method.

DNA intercalating agents such as PMA have previously been used to selectively distinguish between viable and dead bacterial cells. Once in the cell, the dye intercalates with the DNA and, upon photolysis under visible light, produces stable DNA adducts. DNA cross-linked in this way is unavailable for PCR. Environmental samples suspected of containing a mixture of live and dead microbial cells/spores will be treated with PMA, and then incubated in the dark. Thereafter, the sample is exposed to visible light for five minutes, so that the DNA from dead cells will be cross-linked. Following this PMA treatment step, the sample is concentrated by centrifugation and washed (to remove excessive PMA) before DNA is extracted. The 16S rRNA gene fragments will be amplified by PCR to screen the total microbial community using PhyloChip DNA microarray analysis. This approach will detect only the viable microbial community since the PMA intercalated DNA from dead cells would be unavailable for PCR amplification. The total detection time including PCR reaction for low biomass samples will be a few hours.

Numerous markets may use this technology. The food industry uses spore detection to validate new alternative food processing technologies, sterility, and quality. Pharmaceutical and medical equipment companies also detect spores as a marker for sterility. This system can be used for validating sterilization processes, water treatment systems, and in various public health and homeland security applications.

Using Dissimilarity Measures

Gaining the insight you need from microbiome analysis may require some complicated bioinformatics, but it’s entirely manageable. It helps to think ahead about the outcome of your experiment. You’ve likely collected multiple samples from two or more comparison groups, such as diseased and healthy subjects. The subjects will harbor hundreds to thousands of species so you may be already thinking about how to search through the anticipated data barrage. Your subject groups may contain microbiomes with completely distinct families of bacteria, or the subjects may contain many of the same bacterial families. Will there be significant differences in the population levels of certain bacteria? How can we create a visual summary of the results?

We asked an experienced team of experts, Todd DeSantis, Head of Bioinformatics at Second Genome and chief
architect for PhyloChip probe design and analysis algorithms; Justin Kuczynski, Ph.D, Staff Scientist at Second Genome and co-developer of QIIME microbial community analysis software; and Catherine Lozupone, Ph.D, from the University of Colorado at Boulder, co-creator of the UniFrac algorithm and web interfaces, for their insight.

Their number one recommendation: Start with a high-level overview of your experiment by examining how different each sample’s microbiome is from all others and generate a single plot to visualize these differences.

In this article we’ll look at how community-wide dissimilarity measures and ordination techniques can be used to produce just such a high-level overview. We'll first discuss four dissimilarity measures which condense information on thousands of taxa into a single number. Then we’ll inspect an ordination figure where these dissimilarities are plotted and can be interpreted.

Click the link to access the article:
Download "Bioinformatics and the Microbiome: Using Dissimilarity Measures.pdf" Guide

Comparison of Samples Using Dissimilarity Functions

At some point in your analysis you’ll create a boiled-down table. Each row will represent one type of organism such as a genus or species and each column will contain the abundance of that taxon measured in a particular sample. You’ll hear some microbiome jockeys refer to a column as a sample’s “microbial profile” or “community profile”. Although it’s useful to look closely at the abundance of a few particular species of interest across your samples, it is more useful for an overview to synthesize the information on thousands of taxa simultaneously. This is where sample-to-sample dissimilarity measures are powerful. Ecologists use these tools to consider the variation of the entire microbial profile between communities.

Dissimilarity measures deal with just two samples at a time but the result is a matrix of all pair-wise sample dissimilarities. This matrix is much more manageable for creating the overview because statistical techniques such as hierarchical clustering or ordinations can be employed to visualize the important patterns of variation across your samples. The particular dissimilarity measures used may depend on whether you want to assess changes in the abundances of microbes, or merely the incidence (presence/absence) of microbes and whether you prefer to consider information on phylogenetic relationships between taxa or to treat the taxa as unrelated bins. Table 1 displays just four of the multitude of dissimilarity measures available. You could start with these four since they are widely used in the literature and are available in opensource software. You may discover, like our experts, that it’s good practice to use more than one approach to achieve complementary insights.

For the qualitative (presence/absence) approach, both the Sørensen and unweighted Unifrac calculations are popular choices. In considering a pair of samples, the Sørensen index is sensitive to overlap. Specifically it's the ratio of the number of taxa simultaneously detected in both samples compared to the sum of the taxa in both samples. However, while the Sørensen index provides valuable insights, it stops short of using the phylogenetic relationships that exist among all bacteria, something that is offered by unweighted Unifrac calculations. Since bacteria are taxonomically classified into phyla, classes, orders, families, and so on, differences between samples can incorporate the magnitude of the genetic difference between the microbes present in each sample. As a simple example, consider two fecal samples with identical microbial incidence except for two species. Using Unifrac, if those two species are in the same family the pair-wise dissimilarity will be less than if those two species were in distinct families. In cases where phylogenetic relationships between taxa are known, unweighed Unifrac dissimilarity may be preferable over Sørensen dissimilarity.

If you have the ability to produce reliable quantitative data on the abundance of each taxon in your samples, then you’ll have additional dissimilarity choices available. The basic sample-to-sample dissimilarity measurement using the abundance of each taxa is the Bray-Curtis index. The Bray-Curtis function performs a pair-wise normalization by dividing the sum of differences by the sum of all abundances which is helpful when abundance metrics across samples are imperfectly scaled. The index is sensitive to the difference in abundance observed between the same taxa across pairs of samples. To integrate both the relationships between taxa and the abundance fluctuations of those taxa across samples, use the weighted Unifrac measure. For example, since the PhyloChip™ assay tracks abundance changes for thousands of microbes and the genetic distances between those microbes can be measured, weighted Unifrac can aid in resolving subtle variations in composition between microbial communities.

Tip 1:

While applying these measures, one should beware of a common pitfall: If each sample in your experiment was sampled to a different depth, then the sensitivity of detection is inherently unequal and dissimilarity measures may tell you more about your methods than about your samples. If you are using counts of sequences per taxon as your incidence or abundance data, samples with few counts overall will look artificially distinct since many taxa will not be encountered. Keep in mind that you’ll prepare over 1011 gene fragments per specimen and then observe between 102 to 105 by sequencing. Although incidence measures may be more sensitive to this phenomenon, measures that consider abundance are impacted too. It is thus recommended to standardize the number of sequences per sample in your data table before applying dissimilarity measures. Both bar-coded pyrosequencing and Illumina platforms often yield very high variability in the numbers of sequences per sample despite best laboratory efforts in standardizing the amount of DNA per sample. Choosing the optimal number of sequences per sample at which to standardize involves a trade-off between sequencing depth and sample inclusion, since those samples that do not meet the threshold should be excluded. Since the PhyloChip™ assay does not rely on sequence counting techniques, these data adjustments and sample exclusions are not commonly applied.

Ordination – Visual Representation of the Dissimilarities Among Many Samples

Regardless of the dissimilarity measure chosen, the results are usually presented in a visual form in order to summarize the inter-sample relationships. The basic plot seen in many microbial ecology journal articles is a two-dimensional ordination, of which there are several flavors. In ordination plots, each point represents a microbial community sample (see Figure 1) and these points are typically colored by category so the reader can quickly identify which points are, for instance, healthy subjects, disease sub-type 1, disease sub-type 2, etc.

The relationships between two microbial communities can vary in more than two dimensions, for instance if the variation of one subset of the microbial community is primarily driven by disease status, and in a distinct subset is primarily driven by diet, and in yet another distinct subset is primarily driven by gender of the host. Ordination plots are, in a sense, a dimensional reduction technique, analogous to viewing the two dimensional shadow cast by a three dimensional object. However, the two dimensions shown in an ordination plot are an attempt to represent as fully as possible the true multidimensional relationships among the samples. In ordinations, each point represents a single sample or a single microbiome. Points with similar colors belong to the same experimental group. The greater the microbiome dissimilarity between two samples, the further-apart those two points are in the image. Once a plot has been constructed you will then, hopefully, have a good idea of how your samples differ. For example, in Figure 1, it’s clear that the green dots of Group A and the orange dots of Group E are distinct. You may hear microbial ecologists state, “the dissimilarity between the groups is greater than the dissimilarity within groups”. That’s always an interesting observation.

Tip 2:

Once you are comfortable viewing ordinations, here’s a tip for a deeper understanding. The extent to which an ordination was successful in reducing all the comparisons into fewer dimensions is usually noted with numeric results that go along with the ordination plot. For example, the process of constructing a two-dimensional Principal Coordinate Analysis (PCoA) plot will typically report the percent variation explained in each of the two axes (higher numbers mean the two dimensional plot more fully represents the complete variation across the samples). Similarly, a the construction of a two dimensional Non-Metric Multidimensional Scaling (NMDS) plot will typically report “stress”, which is a measure of the variation in the rank order of the true pairwise dissimilarities to the distances shown in the plot (lower values of stress indicate a better representation of the overall variation among samples).
Although the bioinformatics methods applied to microbiome research can be complex, we hope this discussion will
accelerate your ability to get an overview of your data. Since there are many more methods out there to assess and report the differences between communities of microbes, please don’t hesitate to e-mail us questions at
info@secondgenome.com.

 

Whether it’s gut samples or water samples, our goal is to get you closer to applying your discoveries.

Let us know if you need:

• Answers in weeks, not months or years
• Robust, high confidence results through reproducible, standardized assays
• Easy-to-understand synthesis of 16S rRNA gene abundance, rather than complex and time-consuming raw data

The Second Genome Team

 

References:

1. Microbial community resemblance methods differ in their ability to detect biologically relevant patterns. Kuczynski, J., Z. Liu, C. Lozupone, D. McDonald, N. Fierer, and R. Knight (2010) Nat Methods, 7(10), 813-819.
2. Fast UniFrac: facilitating high-throughput phylogenetic analyses of microbial communities including analysis of pyrosequencing and PhyloChip data. Hamady, M., C. Lozupone, and R. Knight, (2010), Isme J, 4(1), 17-27.
3. Species divergence and the measurement of microbial diversity. Lozupone, C.A. and R. Knight, (2008), FEMS Microbiol Reviews. 32(4):557-78.
4. Quantitative and qualitative β diversity measures lead to different insights into factors that structure microbial communities. Lozupone C.A., M. Hamady, S. Kelley, and R. Knight, (2007), Appl Environ Microbiol. 73:1576-1585.

 

Second Genome Announces New SVP of Corporate Development

Second Genome welcomes Mohan S. Iyer to the team. As SVP of Corporate Development, Mohan is responsible for building relationships and working with our partners, including biopharma and nutrition companies, to identify novel biomarkers for classifying patient populations and to assess the impact of novel microbiome-modulating products. Mohan has over 25 years experience in the life sciences industry in a range of functions at Genentech, diaDexus, and most recently as CBO and CFO at Tethys Bioscience. He also brings experience as a management consultant, at The Wilkerson Group, and as an investment banker, at Burrill & Company. He holds an BS in Chemical Engineering from University of Tennessee, an MS in Biomedical Engineering from Duke, and an MBA from Yale University.

Second Genome: Shaping a Better World

Every year the Association of University Technology Managers (AUTM) identifies the world’s most innovative technologies for their Better World Report. This year’s report, “Respond, Recover, Restructure: Technologies Helping the World in the Face of Adversity” features 23 innovations that have changed the way we live and includes our PhyloChip technology in the category of “Technologies to Improve Health” section. The report features the story of how PhyloChip was developed at the Lawrence Berkeley National Laboratory based on the potential of the 16S ribosomal gene and the development of DNA micorarrays and how it took a chance conversation to turn an experimental technology into a commercial one. It also details the power and accuracy of PhyloChip and its wide range of uses including detecting oil-digesting bacteria in the Deepwater Horizon spill, assessing the health of coral reefs, and understanding the human microbiome.

PhyloChip’s worthy companions in the report include a device for analyzing the function of knee joints while they are in motion; the first blood test to diagnose brain injuries; a new high-yield wheat strain; and a small, inexpensive water filter developed from a teabag that can be mass produced and shipped to areas of the world stricken by natural disasters and poverty. “Inclusion in this report re-affirms the importance of microbiome research in finding solutions to medical and environmental problems”, says Todd DeSantis, one of the inventors of PhyloChip. “We know that bacteria live all around us and even within our bodies. Understanding the interplay between the microbiome and the human genome is fundamental to understanding human health. Comparing the gut bacteria of patients suffering from Irritable Bowel Syndrome, for instance, with those in healthy subjects reveals that certain, but not all bacteria, exhibit a population increase with the disease. Bacteroides vulgatus was identified with the PhyloChip as one of those bacteria in a study conducted at Baylor College of Medicine (Saulnier et al., Gastroenterology 2011).”

The Better World Report is part of the Better World Project, which aims to promote public understanding of how new technologies improve quality of life for us all. The report was sponsored by AstraZeneca, Lehigh University, and Massachusetts Institute of Technology among others. Copies of the report are available from betterworldproject.net.

About the author: Ruth Warre is a freelance scientific writer and editor currently living in Toronto. She writes on a variety of subjects from microbiomes to neuroscience, in a variety of mediums from blogs to peer-reviewed articles.