Scalable Tools for Precision Immunology
Our lab aims to advance the emerging field of precision immunology by developing scalable new tools that combine sequencing, synthesis, and big data and applying them to the adaptive immune system.
The adaptive immune system is involved in most (patho)physiological processes and is responsible for preventing, curing, or causing a broad range of diseases. This system functions by relying on a heterogeneous cell population with an enormous diversity of antigen receptors. The specificity of the receptors’ interactions determines the pathophysiology of the immune response. Detailed understanding of such receptor interactions will form the basis of a new generation of ultra-personalized diagnostics/biomarkers and immunotherapies (“precision immunology”). However, we have traditionally been hampered by limitations in our ability to measure and analyze these non-germline encoded receptors interactions.
Three recent trends are coming together to enable the orders-of-magnitude improvements necessary to understand immune interactions. Advances in next- generation sequencing (NGS) have provided a high-throughput data generation capability to any assay that can be encoded in DNA. Meanwhile, rapid improvements in DNA synthesis are enabling massively parallel library screens. Importantly, however, the use of these technologies relies heavily on the effective use of large amounts of data. The recent Hadoop ecosystem of software tools (inspired by Google’s data infrastructure) is enabling the storage and analysis of unprecedented amounts of data, yet it is relatively unknown in the life sciences.
We are interested in applying these technologies broadly, but have started on several specific directions:
Developing methods for rapidly building phage/yeast/ribosome display libraries of interesting antigens. We are using both large arrays of synthesized oligonucleotides and metagenomic approaches to develop reagents that can be used to measure the specificity of query antibodies/TCRs (e.g., using PhIP-seq). We currently have libraries for self antigens and viral antigens, and are creating libraries for gut antigens, food antigens, and allergens. We are also interested creating personal peptide libraries in the context of cancer neopepitopes, for example.
The role of our commensal microbiota in our physiology continues to broaden. Secreted IgA (sIgA) is the most abundantly produced type of antibody and its role in regulating gut homeostasis is becoming increasingly central. We are applying our assays to obtain a high- resolution picture of how the immune system interacts with the microbiota in the context of different diseases. Using germ-free mice, we can analyze the antibody responses to specific microbes. We are also interested in broadly characterizing human sIgA repertoires in the context of disease (e.g., IBD).
As functional genomics assays increasingly rely on larger NGS-based assays, the tools for manipulating these data sets and training machine learning models become increasingly unwieldy. We are developing software libraries for provide functional genomics primitives (e.g., computing genomic intersections) for manipulating large ENCODE-style data sets on computing clusters. These tools will simplify building scalable machine learning models in genomics.