Benjamin Callahan

PreMiEr������������������s microbiome engineering framework will enable the development of a wide range of transformative technologies that solve societal challenges at the interface of health and the environment. However, the dissemination of these same technologies is not without risk as it relies on the responsible development and societal acceptance of microbiome engineering approaches. Thus, in this research core, we will consider the ethical, societal, and policy implications of PreMiEr������������������s evolving microbiome engineering discoveries. There have been national calls for cross-disciplinary and integrated work to better understand the social implications of microbiome science and engineering [1]. In parallel, there is increasing awareness that challenges at the nexus of human- and natural-world coupled systems cannot be solved by technology alone. Through the research and deliberative engagement approaches described below, PreMiEr research will embrace the concept of responsible research and innovation and its elements of anticipation, deliberation, reflexivity, and responsiveness [6]. Particular areas of inquiry will be on social equity of microbiome engineering, ownership and privacy of microbiome data and information, and ethical implications including informed consent, consumer and patient autonomy, beneficence, non-malfeasance, and procedural justice. Core B will also work with the natural scientists and engineers in other thrusts and cores to help identify and address policy and societal questions associated with risk governance and analysis, oversight of microbiome engineering, and equitable distribution of risks and benefits.

The Callahan laboratory develops quantitative and computational methods that improve the precision, accuracy and reproducibility of marker-gene and metagenomics sequencing methods, and implements those methods in open-source and actively supported software. The Callahan laboratory collaborates with other research groups to investigate the role of host-associated microbial communities in various health problems, with a particular focus on the role of the vaginal microbiome preterm birth. Marker-gene and metagenomic sequencing methods have revolutionized the study of the human microbiome, but the relative abundances of microbial taxa measured by these technologies are systematically distorted (or ����������������biased���������������) from the true relative abundances by experiment- and taxa-specific factors we do not understand. Over the next five years we intend to solve the problem of bias in metagenomic sequencing via a multi-part program that includes developing and validating an explanatory mechanistic model of bias in metagenomics experiments, quantifying the sensitivity of downstream analysis methods to metagenomics bias and creating new analysis methods unaffected by bias, and developing open-source software that allows researchers to correct biased marker-gene and metagenomics measurements to their true values. Recently, marker-gene and metagenomic sequencing methods have revealed that the vaginal microbiome of pregnant women can predict risk of preterm birth. However, disagreements between studies on the specific microbial signatures of preterm risk currently limit translation of these results. We intend to develop dense longitudinal profiles with sub-species taxonomic resolution of the vaginal microbiota in hundreds of pregnancies. We will define precise biomarkers of preterm risk as early in possible in pregnancy, and identify candidate taxa that could be targeted as part of therapeutic interventions to reduce the rate of preterm birth. We will synthesize the larger evidence base on this topic while controlling for the different metagenomics biases in different studies in order to identify reproducible biomarkers of preterm birth risk. We will integrate the longitudinal profiles of the vaginal microbiota with other omics measurements of host response to build evidence for potential causative pathways between disturbances in the vaginal microbiota and preterm birth. By developing computational and statistical methods that correct measurements of the microbiome to their true values, the Callahan laboratory seeks to improve the precision and accuracy with which the wider research community can characterize the human microbiome, thereby accelerating our understanding of microbiome- related health conditions and the development of microbiome therapies that improve human health.

Morphological characterization of pollen associated with forensic geologic materials often is used in casework to provide distinct information on sample origin (e.g., Where was the IED built?) and assist with sample-to-sample comparisons. Despite its utility, pollen characterization is not routinely performed given it requires specialized expertise and is very laborious. With advances in sequencing technologies, studies have reported that DNA-based approaches such as DNA metabarcoding could streamline pollen characterization. In this proposed study, we will validate DNA metabarcoding for the characterization of pollen from diverse surface soils. This validation is essential before DNA metabarcoding can be implemented into casework, given previous studies have a) primarily focused on pristine soil samples, which are not representative of evidence material, and b) not completed a side-by-side comparison of the results to those obtained using the traditional morphological approach. To complete the validation, pollen present in ~350 diverse surface soils (i.e., varied climate, geology, pH, depositional history, oxidation state) collected from across the U.S. will be sourced from the U.S. Geological Survey (USGS) archive. For each sample, pollen will be characterized two ways: 1) by targeting two commonly used plant DNA metabarcoding regions (ITS2 and trnL), and 2) using traditional morphological techniques. Statistical analyses will determine whether DNA metabarcoding can be used to both qualitatively (i.e., whether the same plant taxa are identified using both methods) and quantitatively (i.e., whether the relative abundance of identified plant taxa is comparable between the methods) characterize pollen from surface soils. The results from this study would drive forward forensic geology; cost and time effective characterization of pollen from geologic materials using DNA metabarcoding could be implemented into routine casework.

Idiopathic epilepsy is the most common chronic nervous system disorder of dogs. Its cause is poorly understood, but is believed to involve genetic and environmental factors. Treatment with anti-seizure drugs remains the standard of care. However, approximately one-third of dogs fail to achieve satisfactory seizure control, highlighting the need to investigate factors that may influence disease course. An association between epilepsy and inflammatory gastrointestinal disease is well documented in humans, and several other nervous system disorders have been linked to alterations in gut microbial populations, with considerable attention focused on the bacteria Helicobacter and Lactobacilli. The aim of this study is to determine whether dogs with idiopathic epilepsy have alterations in the gastrointestinal environment that might influence disease course. We hypothesize that dogs with idiopathic epilepsy have alterations in the gut microbial population - characterized by the presence of Helicobacter and a decrease in Lactobacillus and resulting inflammation ������������������ that are associated with epilepsy development and outcome. Fecal samples will be obtained from 100 pairs of dogs, consisting of 50 untreated and 50 phenobarbital treated epileptic dogs, and an unaffected dog from the same household. Molecular genetic techniques will be used to evaluate for Helicobacter and Lactobacillus, and samples will be tested for biomarkers of inflammation. Data will be evaluated for associations with treatment outcome. In exploring the association between the gut microbiota and canine epilepsy, this study has the potential to improve our understanding of epilepsy, and ultimately lead to more effective management of this disorder.

Food-borne pathogens enact substantial harms on the American people in the form of illness, lost productivity, and expenses related to mitigation and regulatory compliance. Surveillance and tracing of foodborne pathogens is a key control strategy, but its efficacy is reduced by the longtimes associated with current culture and whole-genome-sequencing approaches. Rapid, accurate and comprehensive pathogen detection would improve the safety and lower the costs of our food supply. We aim to develop a targeted metagenomics methodology that can rapidly (<24 hrs) and precisely identify a broad range of foodborne pathogens from heterogeneous environmental samples. In order to achieve this, we propose to combine the GenomeTrakr and NCBI RefSeq databases with cuttingedge bioinformatics tools developed by the PD that achieve single-nucleotide resolution from amplicon sequencing data of full-length genes to identify E. coli and Salmonella strains to the serovar level (e.g. E. coli O157:H7 or S. enterica Heidelberg). We will validate the resolution and accuracy of this new methodology in silico, on isolates of various pathogenic serovars, and in environmental samples of various types for which pathogen presence and identity were previously established by standard culture-based methods. Our methodology will be distributed to the broader food safety community as open-source and actively-supported software, alongside extensive documentation of its efficacy and best-practices guidance. Successful completion of this project will yield a powerful, usable, and broad-spectrum pathogen surveillance technique that will improve food safety by detecting foodborne pathogens before they reach consumers, and by rapidly tracing outbreaks to their source.

Gardnerella vaginalis has been associated with bacterial vaginosis (BV) and preterm birth (PTB) risk, but with little success in being able to identify, treat, and prevent risk. Increasingly, G. vaginalis variants have been differentially associated with these outcomes and associated pathogenic behaviors. These differences have been linked to vast phylogenetic and phenotypic diversity across the named species, G. vaginalis , which has been suggested to represent as many as 13 genomic species. In order to eventually understand, prevent, and treat pathogenic outcomes, we must define a consistent and reliable method to identify G. vaginalis variants. We have demonstrated the ability to identify the presence and abundance of six subspecies G. vaginalis clades in shotgun metagenomic sequencing data, which is an important step in understanding how these clades may shape their microbial communities. Preliminary evidence suggests potential differences in community signatures associated with each clade. We hypothesize six subspecies G. vaginalis clades differ in overall relative transcription rates, transcription of genes related to pathogenicity, and that fold increases of G. vaginalis transcripts in preterm birth vary among the clades. We also hypothesize that G. vaginalis clades differ in phenotypes that relate to the ability to shape the microbial community and host health. We will use paired metagenomic and metatranscriptomic sequencing data to describe G. vaginalis clade transcription in pregnancy and preterm birth. We will also use in vitro assays to assess epithelial adhesion, biofilm formation, and competitive growth against Lactobacillus spp. of the clades. Successful completion of these aims will demonstrate clinically relevant variation among six G. vaginalis clades. Therefore, these aims represent a unique multidisciplinary approach to the systematic understanding of G. vaginalis diversity, which is necessary for identifying health risks in conditions such as BV and PTB, and targeting successful intervention strategies. These aims will also provide training in both computational and technical skills to foster an independent researcher capable of bridging disciplines to solve complex problems in biomedicine.

We propose to bring the maximum-resolution bioinformatics methods recently developed for bacterial microbiome research to the study of fungal communities via four concrete objectives: (1) Extend the DADA2 method for sequence variant inference to work on variable-length amplicons like those generated from the ITS gene region; (2) Benchmark the accuracy of the extended DADA2 method on Illumina amplicon data from fungal mock communities; (3) Develop a public tutorial for applying DADA2 to fungal amplicon data; (4) Submit a manuscript reporting the benchmarking results to a peer-reviewed open-access journal. The extended DADA2 method we are proposing will provide researchers a more precise and accurate picture of amplicon-sequenced fungal communities than is possible with the current methods in the field.