Tag: Raghuveer Parthasarathy

Using machine learning to classify bacterial species from fluorescent image data

Presenter: Noah Pettinari – Physics

Faculty Mentor(s): Raghuveer Parthasarathy

Session: (In-Person) Oral Panel—Uniquely Their Own

The study of host-microbe interactions has been of growing interest in recent years, with new research highlighting their importance in ecology, human health, developmental biology, and immunology. Fluorescent imaging of larger multispecies bacterial communities within the host microbiome is generally limited to one species per fluorescent channel, greatly limiting the ability to image several species simultaneously. Additionally, the creation and integration of new fluorophores is a slow and labor intensive process, further limiting the use of fluorescent imaging. We assess an algorithm for classifying two bacterial species in vitro within one fluorescent channel using machine learning techniques on morphology data. We then applied this machine learning model to bacterial communities in the rotifer gut, testing new algorithms for removing unwanted autofluorescence along the way.

Building Zebrafish Gut Bacterial Communities From the Bottom Up

Presenter(s): Dylan Martins − Biology

Faculty Mentor(s): Raghuveer Parthasarathy

Poster 44

Research Area: Natural/Physical Science

The intestines of humans and other animals are home to tens of trillions of microbes. These microbial communities play important roles in health and disease, and are composed of dozens to hundreds of interacting species. While the factors that determine a particular species’ presence in the gut are largely unknown, both physical and biochemical interactions between species are likely important. Learning about these factors poses challenges due to the difficulty of performing controlled experiments with existing tools.

This project addresses these challenges by constructing five-species microbial communities in zebrafish, a model vertebrate animal to determine whether these model groupings are stable, and what inter-species interactions are evident. We use zebrafish as a model organism because they can live in a bacterially controlled environment and because their larval transparency allows for live microscopy. Experimentally, we introduce commensal intestinal microbes to larval zebrafish, initially raised germ-free to allow introduction of controlled combinations of bacterial species. Using a combination of conventional dissection and plating assays and three-dimensional live imaging, we have been able to demonstrate the existence of stable multi-species communities, and we can test whether outcomes from two-species competitions contain enough information to allow prediction of multi-species abundances and interactions, of key importance to creating predictive models of the human gut. Further, we find that individual species are differentially sensitive to the presence of other species, and that the community stability is sensitive to the presence of certain species. Correlations can also be identified between species and their spatial structure within the fish gut.

The microbiome is important to health and disease, but it is a complex system which is difficult to understand. By constructing a model system in a vertebrate gut that has an interesting and tractable number of species, we gain insights and reveal principles that might apply to the human microbiome.

Examining Pairwise and Multi-Species Interactions in Larval Zebrafish

Presenter(s): Dylan Martins

Faculty Mentor(s): Raghuveer Parthasarathy

Oral Session 1 O

The microbial communities resident in animal intestines are composed of dozens to hundreds of species and play important roles in host health and disease. The determinants of microbial composition, which may include physical characteristics or biochemical interactions, remain largely unknown. Further, it is unclear for many multi-species consortia whether their species- level makeup can be predicted based on an understanding of pairwise species interactions, or whether higher-order interactions are needed to explain community assembly. It is also unclear how spatial organization plays a role in determining the make up of these complex communities. To address this, we consider commensal intestinal microbes in larval zebrafish, initially raised germ-free to allow introduction of controlled combinations of bacterial species. Using a combination of dissection and plating assays and three-dimensional live imaging, we demonstrate the construction of communities of one to five species and test whether outcomes from two-species competitions contain enough information to predict the abundances in more complex communities. We also quantify changes in species’ spatial distributions induced by the presence of other species, which may explain correlations in their abundances. Lastly, we explore the ability of in vitro interbacterial relationships to predict those of the same bacteria in in vivo association.

Characterizing the relationship between bacterial motility and range expansion

Presenter(s): Noah Pettinari—Physics

Faculty Mentor(s): Raghuveer Parthasarathy

Session 5: To the Moon and Back—Relativity Matters

Self-propelled organisms were first observed under the microscope over 300 years ago . Since then, great strides have been made in characterizing the mechanisms behind motile behavior in bacteria, but current models relating cellular motility to bulk range expansion have not been rigorously tested . To better characterize the relationship between these micro- and macroscale patterns, our research is focused on the analysis of images collected via light sheet fluorescence microscopy of bacterial cells and macroscopic imaging of range expansion . Preliminary results have suggested disagreements between predicted rates of range expansion and cellular motility . Further data and analysis is needed to confirm these results . These findings may highlight the need for the consideration of spatial structure or the possibility of unknown mechanisms in current models .