Mathematical models for unexpected correlations and its applications for biological networks

• Grant number: 13/24516-6
• Support Opportunities: Scholarships abroad - Research
• Effective date (Start): July 01, 2014
• Effective date (End): January 31, 2015
• Field of knowledge: Physical Sciences and Mathematics - Probability and Statistics - Applied Probability and Statistics
• Principal Investigator: Anatoli Iambartsev
• Grantee: Anatoli Iambartsev
• Host Investigator: Andrey Morgun
• Host Institution: Instituto de Matemática e Estatística (IME). Universidade de São Paulo (USP). São Paulo , SP, Brazil
• Research place: Oregon State University (OSU), United States

Abstract

As a result of rapid growth in the field of molecular biology, the recent technological advances enable us to measure the expression of thousands of genes simultaneously. However, simply measuring the expressions of multiple individual genes is insufficient to understand complex biological systems. To relate gene expressions to physiological states and environmental factors, we must utilize gene expression networks. These networks enable identification of previously unknown macroproperties of biological systems that emerge from interactions of multiple genes.Our ability to reveal these emerging properties relies on the precision of network reconstruction from experimental data. The major limitation for network reconstruction is usually the number of samples available for the study. This requirement (large sample size) partially depends on the statistical method used in network reconstruction.We propose a new method for this problem by measuring Proportion of Unexpected Correlations (PUC) that relies on fundamental principles that connect correlation and causality. Based on our preliminary data (the analysis of a limited number of gene expression datasets, the in silico comparison and the mathematical proof), we confirm that the advantage of our approach allows a much greater precision in identifying the proportion of false gene connections attributed to noisy data in a network, than does the FDR. This proposal has the following four components: (i) based on mathematical framework for the PUC that we have established, we will develop a new robust approach for exploring interaction between microbiome and host transcriptome including solutions for different feedback loops and mutual interactions; (ii) we will validate the mathematical approach from (i) on metagenomics data from gut content and intestinal gene expression from mice subject to antibiotics treatment obtained in Prof. Morgun's laboratory.This is an interdisciplinary research. The proposed goals will benefit both the methodology of statistical causal inferences and systems biology approaches for understanding of host-microbe interactions. Also, our approach is expected to become a valuable tool for using global data to approach and to test the next emerging paradigm in biology - hologenome theory of evolution. (AU)