Skip to main content

A Genetic View of Influenza Infection

Periodic Reporting for period 3 - GV-FLU (A Genetic View of Influenza Infection)

Reporting period: 2018-07-01 to 2019-12-31

This project studies individual variation in complex quantitative traits. Our working hypothesis is that multiple immune cell subsets are organized and coordinated in a complex network. Genetic variation in this network mediates the observed physiological disease state. In accordance, our main goal is to develop a novel paradigm that would enable to:
(i) decipher a unified quantitative model for phenotypic diversity that can be used to predict physiological phenotypes from DNA sequence and from immune cell traits
(ii) identify the immune cell traits (e.g. cell subsets, cytokines) that cause inherited variation in immune responses and in organismal phenotypes
(iii) understand how variation in such causal cell traits is encoded by genetic and molecular variation.

Specifically, the project is focused on the immune response to influenza A infection. Influenza A virus is a seasonal epidemics that spreads worldwide and results in up to 500,000 yearly deaths. Influenza A virus causes a wide degree of host phenotypic variation in both human and mouse populations. Host genetics play a major role in determining phenotypic variation during infection, yet a large fraction of this inherited variation cannot not be explained by classical association tools. Therefore, study of Influenza infection in the murine lungs across genetic backgrounds is an attractive model for developing our computational paradigm.

Overall, the proposed project addresses fundamental questions in genetics and immunology: its immediate results will directly impact our understanding of Influenza pathogenesis. The methodologies it establishes will provide a new paradigm for studying the role of immune cells in mediating phenotypic diversity and a paradigm for delivering such a model. To truly make our paradigm a transformative technology, we implement our algorithms as tools that are accessible to the scientific community.
During the first period of the project we have been solving challenging computational problems toward studying the role of immune cells in mediating phenotypic diversity. Specifically we have developed and improved four main algorithms:
(1) POEM - identifying modules of transcribed genes affected by joint (pairwise) genetic effects
(2) VoCAL - identifying genetic variants that are associated with cell type quantities, assuming that these quantities are inferred from bulk gene expression data through deconvolution methods.
(3) DCQ - inferring cell type quantities from Bulk gene expression data. We substantially improved the ability of the algorithm to handle a very large reference datasets.
(4) CoD - A classification method that is tailored for deconvolution-based cell type quantities (this project has been conducted and submitted before the formal onset of the ERC project).

To truly make our paradigm a transformative technology, libraries and source code of all algorithms are freely available. In addition, we implemented the DCQ algorithms in a user friendly tool (the 'ImmQuant' application; available at csgi.tau.ac.il/ImmQuant/) that had become widely used in the scientific community. We have also collected a large part of the influenza-response data that will enable us to identify immune cells that play a role in inter-individual phenotypic variation during the course of in vivo influenza infection.
The progress beyond the state of the art is in three main aspects:
(i) Identifying disease-relevant cell types. Causal models of immune responses have been previously constructed but are not scalable to a large number of cell traits. The methods and paradigm that we developed allow reaching beyond the ‘candidate cell’ approach - where researchers pretend to know the changes in cell types for a particular disease - to an unbiased study in which we may reveal the cell types that play a role in disease biology.
(ii) A paradigm for genome-wide association studies (GWAS) and systems genetic studies. A better understanding of the relationships between variation in genetic information, cell composition and phenotypes, is critically needed. However, current approaches are limited in several aspects. First, GWAS are mainly limited to the relations between genetics and phenotypic diversity. Systems genetic studies extend this notion by adding the molecular network as intermediate layer between the genotype and the phenotype, but yet, inter-individual variation in the cellular composition is largely ignores. In this project we set out to develop a toolbox of computational methodologies that would extend beyond the state-of-the-art standard GWAS and systems genetics approaches, allowing an extended model encompassing variation in genotypes, phenotype, as well as the molecular and cellular network.
(iii) Insights into influenza pathogenesis. Many of the cell subsets that we model are not frequently studied in the context of influenza infection. Thus, this work will allow identifying new functions that play a role in Influenza pathogenesis.

We proceed in three directions: first, we continue developing our computational toolbox toward building quantitative model for phenotypic diversity, as well as identify disease-relevant immune cell traits and understanding the genetic and molecular basis of their variation. Second, we now making our first steps in applying our computational toolbox on our influenza infection data collection. Finally, we continue collecting data from additional mouse individuals to improve the statistical power of predictions. Altogether, we expect to complete these tasks until the end of the project —finalizing our generic framework for studying cell traits and phenotypic diversity, and gaining insights into influenza pathogenesis.