Trey Ideker

Analysis of patient genomes and transcriptomes routinely recognizes new gene sets associated with human disease. Here we present an integrative natural language processing system which infers common functions for a gene set through automatic mining of the scientific literature with biological networks. This system links genes with associated literature phrases and combines these links with protein interactions in a single heterogeneous network. Multiscale functional annotations are inferred based on network distances between phrases and genes and then visualized as an ontology of biological concepts. To evaluate this system, we predict functions for gene sets representing known pathways and find that our approach achieves substantial improvement over the conventional text-mining baseline method. Moreover, our system discovers novel annotations for gene sets or pathways without previously known functions. Two case studies demonstrate how the system is used in discovery of new cancer-related pathways with ontological annotations.

Significance: Protein location and function are dependent on diverse cell states. We develop a conditional function predictor (CoFP) for proteome-wide prediction of condition-specific locations and functions of proteins. In addition to highly accurate retrieval of condition-dependent locations and functions in individual conditions, CoFP successfully discovers dynamic function changes of yeast proteins, including Tsr1, Caf120, Dip5, Skg6, Lte1, and Nnf2, under DNA-damaging stresses. Beyond specific predictions, CoFP reveals a global landscape of changes in protein location and function, highlighting a surprising number of proteins that translocate from the mitochondria to the nucleus or from endoplasmic reticulum to Golgi apparatus under stress. CoFP has the potential to discover previously unidentified condition-specific locations and functions under diverse conditions of cellular growth.

In eukaryotes, many chromatin proteins together regulate gene expression. Chromatin proteins often direct the genomic binding pattern of other chromatin proteins, for example, by recruitment or competition mechanisms. The network of such targeting interactions in chromatin is complex and still poorly understood. Based on genome-wide binding maps, we constructed a Bayesian network model of the targeting interactions among a broad set of 43 chromatin components in Drosophila cells. This model predicts many novel functional relationships. For example, we found that the homologous proteins HP1 and HP1C each target the heterochromatin protein HP3 to distinct sets of genes in a competitive manner. We also discovered a central role for the remodeling factor Brahma in the targeting of several DNA-binding factors, including GAGA factor, JRA, and SU(VAR)3-7. Our network model provides a global view of the targeting interplay among dozens of chromatin components.

Caloric restriction extends lifespan, an effect once thought to involve attenuation of reactive oxygen species (ROS) generated by aerobic metabolism. However, recent evidence suggests that caloric restriction may in fact raise ROS levels, which in turn provides protection from acute doses of oxidant through a process called adaptation. To shed light on the molecular mechanisms of adaptation, we designed a series of genome-wide deletion fitness and mRNA expression screens to identify genes involved in adaptation to hydrogen peroxide. Combined with known transcriptional interactions, the integrated data implicate Yap1 and Skn7 as central transcription factors of both the adaptive and acute oxidative responses. They also identify the transcription factors Mga2 and Rox1 as active exclusively in the adaptive response and show that Mga2 is essential for adaptation. These findings are striking because Mga2 and Rox1 have been thought to control the response to hypoxic, not oxidative, conditions. Expression profiling of mga2Δ and rox1Δ knockouts shows that these factors most strongly regulate targets in ergosterol, fatty-acid, and zinc metabolic pathways. Direct quantitation of ergosterol reveals that its basal concentration indeed depends on Mga2, but that Mga2 is not required for the decrease in ergosterol observed during adaptation. Reactive oxygen species (ROS) damage a variety of structures within the cell, resulting in disease and aging. In a seemingly paradoxical effect termed adaptation, it is possible to prevent damage caused by ROS by pre-treating the cell with a small amount of oxidant. We studied this process in order to identify the mechanisms that provide this protection. Our study identified a number of genes and processes with previously unappreciated roles in adaptation. The mechanisms we identified are remarkable because they are distinct from those previously known to protect the cell from ROS. Although this study is conducted in yeast, the wide conservation of adaptation among many organisms suggests that the results from this study may be widely applicable.

Transcription factors are most commonly thought of as proteins that regulate expression of specific genes, independently of the order of those genes along the chromosome. By screening genome-wide chromatin immunoprecipitation (ChIP) profiles in yeast, we find that more than 10% of DNA-binding transcription factors concentrate at the subtelomeric regions near to chromosome ends. None of the proteins identified were previously implicated in regulation at telomeres, yet genomic and proteomic studies reveal that a subset of factors show many interactions with established telomere binding complexes. For many factors, the subtelomeric binding pattern is dynamic and undergoes flux toward or away from the telomere as physiological conditions shift. We find that subtelomeric binding is dependent on environmental conditions and correlates with the induction of gene expression in response to stress. Taken together, these results underscore the importance of genome structure in understanding the regulatory dynamics of transcriptional networks.

Different experimental designs for protein interaction mapping are modeled to compare their efficiency in completing an interactome map. Testing of the strategy that minimized the final experimental cost in an ongoing Drosophila melanogaster interactome project found 450 high-confidence interactions using only 47 microtiter plates. Comprehensive protein-interaction mapping projects are underway for many model species and humans. A key step in these projects is estimating the time, cost and personnel required for obtaining an accurate and complete map. Here we modeled the cost of interaction-map completion for various experimental designs. We showed that current efforts may require up to 20 independent tests covering each protein pair to approach completion. We explored designs for reducing this cost substantially, including prioritization of protein pairs, probability thresholding and interaction prediction. The best experimental designs lowered cost by fourfold overall and >100-fold in early stages of mapping. We demonstrate the best strategy in an ongoing project in Drosophila melanogaster, in which we mapped 450 high-confidence interactions using 47 microtiter plates, versus thousands of plates expected using current designs. This study provides a framework for assessing the feasibility of interaction mapping projects and for future efforts to increase their efficiency.

The advent of microarray technology has made it possible to classify disease states based on gene expression profiles of patients. Typically, marker genes are selected by measuring the power of their expression profiles to discriminate among patients of different disease states. However, expression-based classification can be challenging in complex diseases due to factors such as cellular heterogeneity within a tissue sample and genetic heterogeneity across patients. A promising technique for coping with these challenges is to incorporate pathway information into the disease classification procedure in order to classify disease based on the activity of entire signaling pathways or protein complexes rather than on the expression levels of individual genes or proteins. We propose a new classification method based on pathway activities inferred for each patient. For each pathway, an activity level is summarized from the gene expression levels of its condition-responsive genes (CORGs), defined as the subset of genes in the pathway whose combined expression delivers optimal discriminative power for the disease phenotype. We show that classifiers using pathway activity achieve better performance than classifiers based on individual gene expression, for both simple and complex case-control studies including differentiation of perturbed from non-perturbed cells and subtyping of several different kinds of cancer. Moreover, the new method outperforms several previous approaches that use a static (i.e., non-conditional) definition of pathways. Within a pathway, the identified CORGs may facilitate the development of better diagnostic markers and the discovery of core alterations in human disease. The advent of microarray technology has drawn immense interest to identify gene expression levels that can serve as biomarkers for disease. Marker genes are selected by examining each individual gene to see how well its expression level discriminates different disease types. In complex diseases such as cancer, good marker genes can be hard to find due to cellular heterogeneity within the tissue and genetic heterogeneity across patients. A promising technique for addressing these challenges is to incorporate biological pathway information into the marker identification procedure, permitting disease classification based on the activity of entire pathways rather than simply on the expression levels of individual genes. However, previous pathway-based methods have not significantly outperformed gene-based methods. Here, we propose a new pathway-based classification procedure in which markers are encoded not as individual genes, nor as the set of genes making up a known pathway, but as subsets of “condition-responsive genes (CORGs)” within those pathways. Using expression profiles from seven different microarray studies, we show that the accuracy of this method is significantly better than both the conventional gene- and pathway- based diagnostics. Furthermore, the identified CORGs may facilitate the development of effective diagnostic markers and the discovery of molecular mechanisms underlying disease.

An epistasis map (E-MAP) was constructed in the fission yeast, Schizosaccharomyces pombe , by systematically measuring the phenotypes associated with pairs of mutations. This high-density, quantitative genetic interaction map focused on various aspects of chromosome function, including transcription regulation and DNA repair/replication. The E-MAP uncovered a previously unidentified component of the RNA interference (RNAi) machinery ( rsh1 ) and linked the RNAi pathway to several other biological processes. Comparison of the S. pombe E-MAP to an analogous genetic map from the budding yeast revealed that, whereas negative interactions were conserved between genes involved in similar biological processes, positive interactions and overall genetic profiles between pairs of genes coding for physically associated proteins were even more conserved. Hence, conservation occurs at the level of the functional module (protein complex), but the genetic cross talk between modules can differ substantially.