Causal Inference and the Construction of Predictive Network Models in Biology

摘要: With the digital universe now having surpassed zetabyte threshold, push is on to expand advanced high-performance computing infrastructures manage and store this vast in ways that facilitate mining of data, then developing applying more sophisticated mathematical algorithms extract knowledge from it. The promise effectively big data nothing less than achieving a higher level understanding nearly every facet life, climate change complex patterns financial markets, complexity living systems. life sciences stand poised lead both generation realization dramatic benefit it, whether predicting preventing next outbreak or uncovering best treat, prevent, cure common human diseases. We can score variations DNA across whole genomes; RNA levels alternative isoforms, metabolite levels, protein state information transcriptome, metabolome proteome; methylation status methylome; construct extensive protein–protein protein–DNA interaction maps, all comprehensive fashion at scale populations individuals. Interactions among these molecular entities define web biological processes give rise higher-order phenotypes, including disease. development analytical approaches simultaneously integrate different dimensions essential if we are meaning large-scale elucidate In chapter I describe number aimed inferring causal relationships variables very datasets by leveraging variation as systematic perturbation source. inference procedures also demonstrated enhance ability reconstruct truly predictive, probabilistic gene networks reflect underlying phenotypes such By integrating many networks, detail examples how apply network models uncover like expression between high-order traits