Behind the Data with Dr. Emily Hollister

05.05.20

In a new series profiling our team of microbiome experts, we have set out to understand microbiome research by asking our bioinformaticians – who have published highly cited papers ranging from topics on paleomicrobiology to semi-industrial profiling of the microbiome. Our bioinformaticians are experts in the human microbiome at birth and in disease states. They have even discovered the major impacts of immigration on the microbiome. At Diversigen, our bioinformatics team is ready to help plan your project and help you accelerate your microbiome discoveries.

Dr. Hollister, the Director of R&D, Computational Biology at Diversigen, offered to start our new series.

Over the course of her academic career and her time with Diversigen, Dr. Emily Hollister has developed microbiome bioinformatics procedures for biofuel reactor systems, childhood health, and the environmental microbiome. Researching core functions of the microbiome, Dr. Hollister has over 70 publications and helped develop Mothur, an early microbiome-focused bioinformatics tools, which has since been cited over 13,000 times.

You worked on Mothur, one of the earliest microbiome bioinformatics tools. How has bioinformatics in the microbiome changed since then, and how has Diversigen advanced bioinformatic tools?

Dr. Hollister: “Mothur, one of the first microbiome bioinformatics platforms – developed specifically for analyzing data from amplicon sequencing – was created by Dr. Patrick Schloss and his research team who were then based at the University of Massachusetts. During early development, Dr. Schloss reached out to the user community to collaborate on Mothur development. I was able to participate in that process while also using Mothur in some of my earliest microbiome manuscripts.

Microbiome analysis was streamlined by Mothur by packaging improved bioinformatic tools into a platform that researchers could readily access to analyze their data. Before Mothur, 16S and ITS amplicon sequence analysis required a high-powered server or even a supercomputer to process early Illumina and 454 data sets, which seemed like huge amounts of data at the time! The same work that required a university supercomputer then, can now be processed at home on my laptop in a matter of hours.

The next-generation sequencing methods that were being introduced to the field required new bioinformatic algorithms and software. In addition to Mothur, QIIME and a variety of other tools and platforms were created to solve these challenges for amplicon data, and alternative platforms became available for whole genome shotgun sequence data.

At Diversigen we’ve developed a full set of bioinformatics pipelines for amplicon and whole genome shotgun sequencing, and their results are packaged it into a highly accessible and user-friendly online report -the CoreAnalysis Report.”

As a professor at Baylor College of Medicine, you worked in microbial ecology and bioinformatics at Texas Children’s Hospital. Could you tell us more about your research into the developing pediatric microbiome?

Dr. Hollister: “After the Human Microbiome Project established parameters describing the healthy adult human microbiome, it was apparent that human microbiome research was a wide-open field. There were so many areas to study. Researchers began adapting the tools and lessons learned from the HMP to look at the pediatric microbiome, as well as ways in which the human microbiome varied in the context of health and disease. Studies have established that the pediatric microbiome develops over time, and animal models suggest that there may be key windows for development and intervention. In the case of lifestyle diseases like metabolic syndrome or obesity, exposure risks in childhood can carry forward to adulthood, and it is believed that by better understanding the pediatric microbiome we may delineate opportunities for intervention or redirection to promote better health outcomes later in life.

The field has learned that the pediatric microbiome can be impacted by many variables, including birth mode, diet, environment, and exposure to antibiotics or other medications. These factors contribute to differences between individuals and can represent confounders that make comparisons between individuals and at different points in the development timeline potentially difficult. We’ve learned that careful study design, well-powered studies, thoughtful consideration of clinical and lifestyle factors, and the utilization of this information can be key to making meaningful conclusions.”

You have also researched the human gut microbiome and wrote, Compositional and functional features of the gastrointestinal microbiome and their effects on human health, which defines the GI microbiome in health. Could you tell us more about how your early work influenced the development of microbiome services at Diversigen?

Dr. Hollister: “Compositional and functional features of the gastrointestinal microbiome and their effects on human health” was meant to be a review of the state of our knowledge regarding the gastrointestinal microbiome (at least when it was written). Microbiome studies conducted at various sites along the GI tract have demonstrated site-specific variation. Factors like pH, oxygen levels, nearby organs, and diet, age, and relative health all impact the profile of the microbiome.

The many variables affecting the gastrointestinal microbiome have demonstrated to the field the challenges associated with designing experimental microbiome studies – many factors needed to be considered or controlled to be able to draw a meaningful conclusion.

When designing a study with Diversigen, our team draws upon their own microbiome research experiences to help other researchers plan and prepare for potential confounding factors and biases.

Diversigen has a team of bioinformaticians and researchers with backgrounds in:

gastrointestinal microbiome,

paleomicrobiology,
the environmental microbiome,
the virome, and the
mycobiome.

We have seen firsthand the power of influencing the small things from the start. Having learned from our collective past experiences and being able to ask the right questions regarding issues like study design, collection strategy, and controls benefits any project by impacting the quality of data generated downstream and ultimately helping draw more meaningful conclusions.”

As part of a meta-analysis, you looked at preterm infants with intestinal dysbiosis. Can you tell us more about what your research found?

Dr. Hollister: “Collaborating with physician scientists at Texas Children’s Hospital and others, we asked to what degree common trends could be found among observations of preterm infant microbiomes associated necrotizing enterocolitis – a devastating disease that affects preterm infants.

Necrotizing enterocolitis can cause long term health effects and even death. There are currently no early diagnostic or prognostic tests for it, and treatment is largely empiric. Researchers looking to better understand the disease and identify novel therapies and improved strategies for patient management, examined the body of literature available for similar signals and hoped for novel discoveries in the microbiome.

Comparing microbiome studies in a meta-analysis can be difficult if the methods used vary substantially between groups. Extraction method, stabilization, sequencing approach, and the bioinformatics used can all have downstream effects on the study result – interstudy variability can occur from differences at any point of the experiment from sample collection to analysis.

This meta-analysis found some signals that were consistent across the published literature, however, there was also a tremendous amount of background noise which limited the number of associations we were able to find between the microbiome and necrotizing enterocolitis. Researchers in the microbiome field now recognize the issue of noise across studies, and many have begun to push for greater standardization of procedures to facilitate the re-use of data and cross comparison studies in the future.

The efficiency of many bio-industrial applications depends on the key taxonomic groups inside the microbiome. What are the major differences between analyzing microbes involved in a bio fermentation process versus the gut microbiome?

Dr. Hollister: “They are more similar than they might appear to be on the surface. In my past research, I studied biofuel reactor systems, and the same fundamental processes – i.e., consumption of carbohydrates, harvest of energy from them, and production of short chain fatty acids – take place in industrial systems like these and the gut microbiome. Despite utilizing similar fundamental processes, different considerations are needed when designing an experiment for one particular system over another.

The species inhabiting an industrial microbiome can be fairly unique, because industrial microbiomes usually have a desired objective or end-product that must be produced at scale and often requires that its microbes can thrive under harsh conditions. Researchers may go to extreme environments like hypersaline lakes or thermal vents, participating in what is known as bio-prospecting, to search for microbes that excel at surviving under extreme heat or pressure or in toxic environments so that they can be utilized in industrial processes.

Biofermentation utilizing microbes to metabolize substrates into specific products is a growing application. You have profiled the biodiversity of semi-industrial facilities utilizing bio fermentation procedures – what can you tell us about the industrial microbiome?

Dr. Hollister: “The efficiency of certain microbes — alone or as a group — to produce specific metabolites of interest is critical in industrial manufacturing. Microbes can be used to produce products ranging from biofuels to therapeutic small molecules. The Mix-Alco platform, which I previously worked on, is a biofuel production approach which leverages a ‘dirty’ starter mix to generate mixed alcohols from organic materials (e.g., crop waste). ‘Dirty’ mixes refer to a mixed, and potentially undefined, community of microbes versus the use a single strain. Carbohydrates are consumed and fermented by these microbes to generate alcohols which can be converted into fuels. Industrial applications of the microbiome are unique and ever-changing. Whether producing alcohols or growing batches of probiotic microbes at scale, the microbes involved are under constant evolutionary pressure and may undergo genetic change. Under normal circumstances, these changes are minimal and result in no change in function. However, it is a possible that certain mutations can occur which result in advantages to the industrial process. This is where genome sequencing can provide information regarding genetic drift and potential change in function. Microbes being managed at industrial scale are also at risk of contamination by other microbes. This can lead to changes in composition and can impact the way in which a process functions, making it important to profile the broader microbial community for changes on a fairly regular basis.

The virome – which you recently presented on in a webinar titled, “Viral Metagenomics: Identifying needles in the haystack and unlocking opportunities for discovery” – is a growing area of research. What is Diversigen doing now to support viral metagenomic research?

Dr. Hollister: “Researchers have been studying the virome for several decades now, but newer research demonstrating the importance of viruses in microbiome-mediated conditions, as well as viral surveillance studies (e.g., in light of the novel Coronavirus (SARS-COV-2)) have attracted new interest to the field. Many researchers are shifting their focus deeper into the microbiome to target viruses that inhabit bacterial communities. Researching bacteria in the microbiome is typically easier because of their larger size, greater abundance, and more readily available tools, whereas the virome is challenging due to the smaller size, lack of universal marker genes, and incomplete reference databases. Bacteria-focused studies have come up short in identifying answers for some conditions, and many are hopeful that the virome will lead to new discoveries and novel understanding of the microbiome.

Diversigen has created a viral metagenomic research platform available to support virome research. Our pipeline can sequence the viral metagenome for both DNA and RNA viruses simultaneously in simple or complex samples.”

Available to every research project conducted by Diversigen is their Core Plus report. What goes into providing microbiome insights to researchers within this report?

Dr. Hollister: “The Core Plus report includes a comprehensive study report and study data helping generating insights in microbiome studies. Key to making novel insights is also putting study results in context of all the other factors. Considering these factors adds layers of information that help generate deeper insights. Whether a case control study, an intervention or a timeline, many variables can impact end results. When observing significant differences between, for example, treatment A and treatment B, we can then ask what the major factors driving those changes are: are there lifestyle differences in the organisms, are the metabolisms different, what are the functions, did the abundance of species shift and was this from a treatment or intervention? Placing everything under careful consideration of how and why it is occurring helps create the meaningful conclusions that lead to novel new discoveries.

At Diversigen, we can also help researchers by putting their research into the perspective of the current body of literature. Having a diverse team of experts means we have bioinformaticians and data scientists who can draw upon their own publications and expertise to make informed recommendations regarding study design, data analysis, and interpretation.”

Initially, your research began in the soil microbiome and has transitioned to the virome. What are the benefits to having expertise in multiple microbiome fields?

Dr. Hollister: “Although the microbiome as a field relies on a common set of tools and approaches, each sub-discipline has had to adapt these to their own sample types and conditions. For example, the presence of substances that can inhibit molecular reagents varies from sample type to sample type and will impact DNA extraction, and whether or not a high degree of host burden (or by-catch) is present will impact downstream analysis methods. Given the range of sample types that can be encountered in environmental sampling and the challenges associated with them, researchers have learned that lessons and tricks from environmental sequencing can sometimes be applied to help other processes. When researching across areas, one becomes flexible in their methods to control for variables and accepts that there could be unknowns in their data. While human-associated samples are often well-represented in reference databases, it is possible that samples from livestock, animal models, or environmental samples may have a substantial proportion of their sequencing reads that do not align with any known reference. Having a range of knowledge across microbiome sub-fields allows our experts to consider the reasons for why there may be unknowns in the data. This could be poor representation of our reference databases, sometimes referred to as microbial dark matter. As researchers leverage shotgun metagenomic assemblies to access unidentified genomes, new organisms are being identified, and we’re discovering novel insights about them that one day could develop into therapies or treatments.

Influencing the early stages of study design maximizes downstream results and having expertise across multiple microbiome fields helps to design studies and develop robust bioinformatic tools to analyze and control for confounding factors.”

What are the future research goals of Diversigen?

Dr. Hollister: “Diversigen’s R&D teams are developing new services and products in the wet lab, as well as establishing new protocols for sequence analysis.

We see multi-omics as the future of our field, and we seek to establish a variety of ways to analyze multi–omic information as researchers expand the amounts of data they generate in any given experiment.

Just recently, we announced a collaboration with Alimentive (formerly known as Robarts) regarding a clinical trial for patients with severe ulcerative colitis [SUC]. We are working together to layer microbiome and other ‘omics data with deep clinical phenotyping of individuals to provide deeper understanding of this disease and hopefully identify new options for patient treatment and disease management.”

Dr. Emily Hollister worked on the first generation of bioinformatics tools to analyze the microbiome and has since published research to elucidate disease in children’s gut microbiome. Her research has powered microbiome discovery and helped establish Diversigen’s metagenomics pipelines. Environmental microbiome research impacts the gut microbiome which in turn impacts research on the human virome. Diversigen is developing tools for each niche of the microbiome to help accelerate microbiome discovery.

Ready to talk to Diversigen microbiome experts like Dr. Hollister? Our bioinformaticians are ready to help you plan your project. Start the discussion at info@diversigen.com

Subscribe to our blog and stay tuned for the our next featured microbiome expert – Tonya Ward (link to Tonya’s profile when posted) – the director of data science & bioinformatics at Diversigen.

The information presented in this blog is made available for general information purposes and may include content provided by third parties. We do not warrant the accuracy, completeness or usefulness of this information and disclaim all liability arising from reliance on it. The statements and opinions expressed in these materials, other than the content provided by us, are solely those of the person or entity making the statements.