Search

Breast cancer comprises a group of distinct subtypes that despite having similar histologic appearances, have very different metastatic potentials. Being able to identify the biological driving force, even for a subset of patients, is crucially important given the large population of women diagnosed with breast cancer. Here, we show that within a subset of patients characterized by relatively high estrogen receptor expression for their age, the occurrence of metastases is strongly predicted by a homogeneous gene expression pattern almost entirely consisting of cell cycle genes (5-year odds ratio of metastasis, 24.0; 95% confidence interval, 6.0-95.5). Overexpression of this set of genes is clearly associated with an extremely poor outcome, with the 10-year metastasis-free probability being only 24% for the poor group, compared with 85% for the good group. In contrast, this gene expression pattern is much less correlated with the outcome in other patient subpopulations. The methods described here also illustrate the value of combining clinical variables, biological insight, and machine-learning to dissect biological complexity. Our work presented here may contribute a crucial step towards rational design of personalized treatment.

Supplementary Tables S1-S4 from A Cell Proliferation Signature Is a Marker of Extremely Poor Outcome in a Subpopulation of Breast Cancer Patients

Supplementary Figures S1-S10 from A Cell Proliferation Signature Is a Marker of Extremely Poor Outcome in a Subpopulation of Breast Cancer Patients

Methods, Figure legends

In response to the unmet need for timely accurate diagnosis and prognosis of acute infections and sepsis, host-immune-response-based tests are being developed to help clinicians make more informed decisions including prescribing antimicrobials, ordering additional diagnostics, and assigning level of care. One such test (InSep™, Inflammatix, Inc.) uses a 29-mRNA panel to determine the likelihood of bacterial infection, the separate likelihood of viral infection, and the risk of physiologic decompensation (severity of illness). The test, being implemented in a rapid point-of-care platform with a turnaround time of 30 min, enables accurate and rapid diagnostic use at the point of impact. In this report, we provide details on how the 29-biomarker signature was chosen and optimized, together with its molecular, immunological, and medical significance to better understand the pathophysiological relevance of altered gene expression in disease. We synthesize key results obtained from gene-level functional annotations, geneset-level enrichment analysis, pathway-level analysis, and gene-network-level upstream regulator analysis. Emerging findings are summarized as hallmarks on immune cell interaction, inflammatory mediators, cellular metabolism and homeostasis, immune receptors, intracellular signaling and antiviral response, and converging themes on neutrophil degranulation and activation involved in immune response, interferon, and other signaling pathways.

A new safety testing paradigm that relies on gene expression biomarker panels was developed to easily and quickly identify drug-induced injuries across tissues in rats prior to drug candidate selection. Here, we describe the development, qualification, and implementation of gene expression signatures that diagnose tissue degeneration/necrosis for use in early rat safety studies. Approximately 400 differentially expressed genes were first identified that were consistently regulated across 4 prioritized tissues (liver, kidney, heart, and skeletal muscle), following injuries induced by known toxicants. Hundred of these “universal” genes were chosen for quantitative PCR, and the most consistent and robustly responding transcripts selected, resulting in a final 22-gene set from which unique sets of 12 genes were chosen as optimal for each tissue. The approach was extended across 4 additional tissues (pancreas, gastrointestinal tract, bladder, and testes) where toxicities are less common. Mathematical algorithms were generated to convert each tissue’s 12-gene expression values to a single metric, scaled between 0 and 1, and a positive threshold set. For liver, kidney, heart, and skeletal muscle, this was established using a training set of 22 compounds and performance determined by testing a set of approximately 100 additional compounds, resulting in 74%–94% sensitivity and 94%–100% specificity for liver, kidney, and skeletal muscle, and 54%–62% sensitivity and 95%–98% specificity for heart. Similar performance was observed across a set of 15 studies for pancreas, gastrointestinal tract, bladder, and testes. Bundled together, we have incorporated these tissue signatures into a 4-day rat study, providing a rapid assessment of commonly seen compound liabilities to guide selection of lead candidates without the necessity to perform time-consuming histopathologic analyses.

Summary The pandemic 2019 novel coronavirus disease (COVID-19) shares certain clinical characteristics with other acute viral infections. We studied the whole-blood transcriptomic host response to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) using RNAseq from 24 healthy controls and 62 prospectively enrolled patients with COVID-19. We then compared these data to non-COVID-19 viral infections, curated from 23 independent studies profiling 1,855 blood samples covering six viruses (influenza, respiratory syncytial virus (RSV), human rhinovirus (HRV), severe acute respiratory syndrome coronavirus 1 (SARS-CoV-1), Ebola, dengue). We show gene expression changes in COVID-19 versus non-COVID-19 viral infections are highly correlated (r = 0.74, p < 0.001). However, we also found 416 genes specific to COVID-19. Inspection of top genes revealed dynamic immune evasion and counter host responses specific to COVID-19. Statistical deconvolution of cell proportions maps many cell type proportions concordantly shifting. Discordantly increased in COVID-19 were CD56bright natural killer cells and M2 macrophages. The concordant and discordant responses mapped out here provide a window to explore the pathophysiology of the host response to SARS-CoV-2., Graphical Abstract, Highlights • Whole blood transcriptomics were generated via RNAseq for 62 COVID-19 patients • Curated 23 whole blood transcriptomic studies (1855 samples) of non-COVID-19 viral infections • Discovered 416 COVID-19-specific genes, despite overall correlation with non-COVID-19 • Revealed subset of immune cell proportions discordantly shifted in COVID-19 infections, Molecular Biology; Immunology: Bioinformatics; Transcriptomics

Objectives Complex critical syndromes like sepsis and coronavirus disease 2019 may be composed of underling "endotypes," which may respond differently to treatment. The aim of this study was to test whether a previously defined bacterial sepsis endotypes classifier recapitulates the same clinical and immunological endotypes in coronavirus disease 2019. Design Prospective single-center observational cohort study. Setting Patients were enrolled in Athens, Greece, and blood was shipped to Inflammatix (Burlingame, CA) for analysis. Patients Adult patients within 24 hours of hospital admission with coronavirus disease 2019 confirmed by polymerase chain reaction and chest radiography. Interventions None. Measurements and main results We studied 97 patients with coronavirus disease 2019, of which 50 went on to severe respiratory failure (SRF) and 16 died. We applied a previously defined 33-messenger RNA classifier to assign endotype (Inflammopathic, Adaptive, or Coagulopathic) to each patient. We tested endotype status against other clinical parameters including laboratory values, severity scores, and outcomes. Patients were assigned as Inflammopathic (29%), Adaptive (44%), or Coagulopathic (27%), similar to our prior study in bacterial sepsis. Adaptive patients had lower rates of SRF and no deaths. Coagulopathic and Inflammopathic endotypes had 42% and 18% mortality rates, respectively. The Coagulopathic group showed highest d-dimers, and the Inflammopathic group showed highest C-reactive protein and interleukin-6 levels. Conclusions Our predefined 33-messenger RNA endotypes classifier recapitulated immune phenotypes in viral sepsis (coronavirus disease 2019) despite its prior training and validation only in bacterial sepsis. Further work should focus on continued validation of the endotypes and their interaction with immunomodulatory therapy.

SARS-CoV-2 pandemic, the fourth pandemic of the decade, has underscored gaps in global pandemic preparedness and the need for generalizable tests to avert overwhelming healthcare systems worldwide, irrespective of a virus. We integrated 4,780 blood transcriptome profiles from patients infected with one of 16 viruses across 34 independent cohorts from 18 countries, and 71 scRNA-seq profiles of 264,224 immune cells across three independent cohorts. We found a myeloid cell-dominated conserved host response associated with severity. It showed increased hematopoiesis, myelopoiesis, and myeloid-derived suppressor cells with increased severity. We identified four gene modules that delineate distinct trajectories associated with mild and severe outcomes, and show the interferon response was decoupled from protective host response during severe viral infection. These modules distinguished non-severe from severe viral infection with clinically useful accuracy. Together, our findings provide insights into immune response dynamics during viral infection, and identify factors that may influence patient outcomes.

Romosozumab (EVENITY™ [romosozumab-aqqg in the US]) is a humanized monoclonal antibody that inhibits sclerostin and has been approved in several countries for the treatment of osteoporosis in postmenopausal women at high risk of fracture. Sclerostin is expressed in bone and aortic vascular smooth muscle (AVSM). Its function in AVSM is unclear but it has been proposed to inhibit vascular calcification, atheroprogression, and inflammation. An increased incidence of positively adjudicated serious cardiovascular adverse events driven by an increase in myocardial infarction and stroke was observed in romosozumab-treated subjects in a clinical trial comparing alendronate with romosozumab (ARCH; NCT01631214) but not in a placebo-controlled trial (FRAME; NCT01575834). To investigate the effects of sclerostin inhibition with sclerostin antibody on the cardiovascular system, a comprehensive nonclinical toxicology package with additional cardiovascular studies was conducted. Although pharmacodynamic effects were observed in the bone, there were no functional, morphological, or transcriptional effects on the cardiovascular system in animal models in the presence or absence of atherosclerosis. These nonclinical studies did not identify evidence that proves the association between sclerostin inhibition and adverse cardiovascular function, increased cardiovascular calcification, and atheroprogression.

Background: Anti-TNF-alpha (anti-TNFα) therapies have transformed the care and management of inflammatory bowel disease (IBD). However, they are expensive and ineffective in greater than 50% of patients, and they increase the risk of infections, liver issues, arthritis, and lymphoma. With 1.6 million Americans suffering from IBD and global prevalence on the rise, there is a critical unmet need in the use of anti-TNFα therapies: a test for the likelihood of therapy response. Here, as a proof-of-concept, we present a multi-mRNA signature for predicting response to anti-TNFα treatment to improve the efficacy and cost-to-benefit ratio of these biologics. Methods: We surveyed public data repositories and curated four transcriptomic datasets (n = 136) from colonic and ileal mucosal biopsies of IBD patients (pretreatment) who were subjected to anti-TNFα therapy and subsequently adjudicated for response. We applied a multicohort analysis with a leave-one-study-out (LOSO) approach, MetaIntegrator, to identify significant differentially expressed (DE) genes between responders and non-responders and then used a greedy forward search to identify a parsimonious gene signature. We then calculated an anti-TNFα response (ATR) score based on this parsimonious gene signature to predict responder status and assessed discriminatory performance via an area-under-receiver operating-characteristic curve (AUROC). Results: We identified 324 significant DE genes between responders and non-responders. The greedy forward search yielded seven genes that robustly distinguish anti-TNFα responders from non-responders, with an AUROC of 0.88 (95% CI: 0.70–1). The Youden index yielded a mean sensitivity of 91%, mean specificity of 76%, and mean accuracy of 86%. Conclusions: Our findings suggest that there is a robust transcriptomic signature for predicting anti-TNFα response in mucosal biopsies from IBD patients prior to treatment initiation. This seven-gene signature should be further investigated for its potential to be translated into a predictive test for clinical use.

Drug induced liver injury (DILI) is one of the key safety concerns in drug development. To assess the likelihood of drug candidates with potential adverse reactions of liver, we propose a compound attributes-based approach to predicting hepatobiliary disorders that are routinely reported to US Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS). Specifically, we developed a support vector machine (SVM) model with recursive feature extraction, based on physicochemical and structural properties of compounds as model input. Cross validation demonstrates that the predictive model has a robust performance with averaged 70% of both sensitivity and specificity over 500 trials. An independent validation was performed on public benchmark drugs and the results suggest potential utility of our model for identifying safety alerts. This in silico approach, upon further validation, would ultimately be implemented, together with other in vitro safety assays, for screening compounds early in drug development.

Genome instability is a hallmark of most human cancers and is exacerbated following replication stress. However, the effects that drugs/xenobiotics have in promoting genome instability including chromosomal structural rearrangements in normal cells are not currently assessed in the genetic toxicology battery. Here, we show that drug-induced replication stress leads to increased genome instability in vitro using proliferating primary human cells as well as in vivo in rat bone marrow (BM) and duodenum (DD). p53-binding protein 1 (53BP1, biomarker of DNA damage repair) nuclear bodies were increased in a dose-dependent manner in normal proliferating human mammary epithelial fibroblasts following treatment with compounds traditionally classified as either genotoxic (hydralazine) and nongenotoxic (low-dose aphidicolin, duvelisib, idelalisib, and amiodarone). Comparatively, no increases in 53BP1 nuclear bodies were observed in nonproliferating cells. Negative control compounds (mannitol, alosteron, diclofenac, and zonisamide) not associated with cancer risk did not induce 53BP1 nuclear bodies in any cell type. Finally, we studied the in vivo genomic consequences of drug-induced replication stress in rats treated with 10 mg/kg of cyclophosphamide for up to 14 days followed by polymerase chain reaction-free whole genome sequencing (30X coverage) of BM and DD cells. Cyclophosphamide induced chromosomal structural rearrangements at an average of 90 genes, including 40 interchromosomal/intrachromosomal translocations, within 2 days of treatment. Collectively, these data demonstrate that this drug-induced genome instability test (DiGIT) can reveal potential adverse effects of drugs not otherwise informed by standard genetic toxicology testing batteries. These efforts are aligned with the food and drug administration's (FDA's) predictive toxicology roadmap initiative.

Mapping network analysis in cells and tissues can provide insights into metabolic adaptations to changes in external environment, pathological conditions, and nutrient deprivation. Here, we reconstructed a genome-scale metabolic network of the rat liver that will allow for exploration of systems-level physiology. The resulting in silico model (iRatLiver) contains 1,882 reactions, 1,448 metabolites, and 994 metabolic genes. We then used this model to characterize the response of the liver’s energy metabolism to a controlled perturbation in diet. Transcriptomics data were collected from the livers of Sprague Dawley rats at 4 or 14 days of being subjected to 15%, 30%, or 60% diet restriction. These data were integrated with the iRatLiver model to generate condition-specific metabolic models, allowing us to explore network differences under each condition. We observed different pathway usage between early and late time points. Network analysis identified several highly connected “hub” genes (Pklr, Hadha, Tkt, Pgm1, Tpi1, and Eno3) that showed differing trends between early and late time points. Taken together, our results suggest that the liver’s response varied with short- and long-term diet restriction. More broadly, we anticipate that the iRatLiver model can be exploited further to study metabolic changes in the liver under other conditions such as drug treatment, infection, and disease.

Viral infections induce a conserved host response distinct from bacterial infections. We hypothesized that the conserved response is associated with disease severity and is distinct between patients with different outcomes. To test this, we integrated 4,780 blood transcriptome profiles from patients aged 0 to 90 years infected with one of 16 viruses, including SARS-CoV-2, Ebola, chikungunya, and influenza, across 34 cohorts from 18 countries, and single-cell RNA sequencing profiles of 702,970 immune cells from 289 samples across three cohorts. Severe viral infection was associated with increased hematopoiesis, myelopoiesis, and myeloid-derived suppressor cells. We identified protective and detrimental gene modules that defined distinct trajectories associated with mild versus severe outcomes. The interferon response was decoupled from the protective host response in patients with severe outcomes. These findings were consistent, irrespective of age and virus, and provide insights to accelerate the development of diagnostics and host-directed therapies to improve global pandemic preparedness., Graphical abstract, Viral infections induce a conserved host response distinct from bacterial infections, but whether this conserved response distinguishes severity is unclear. Zheng et al. analyzed >5000 bulk and single-cell transcriptome profiles from patients infected with one of 16 viruses, including SARS-CoV-2, Ebola, and chikungunya. They identified protective and detrimental host response modules that distinguish patients with mild or severe outcomes.

Inhibition of sclerostin with sclerostin antibody (Scl-Ab) has been shown to stimulate bone formation, decrease bone resorption, and increase bone mass in both animals and humans. To obtain insight into the temporal cellular and transcriptional changes in the osteoblast (OB) lineage associated with long-term Scl-Ab treatment, stereological and transcriptional analyses of the OB lineage were performed on lumbar vertebrae from aged ovariectomized rats. Animals were administered Scl-Ab 3 or 50mg/kg/wk or vehicle (VEH) for up to 26weeks (d183), followed by a treatment-free period (TFP). At 50mg/kg/wk, bone volume (BV/total volume [TV]) increased through d183 and declined during the TFP. Bone formation rate (BFR/bone surface [BS]) and total OB number increased through d29, then progressively declined, coincident with a decrease in total osteoprogenitor (OP) numbers from d29 through d183. Analysis of differentially expressed genes (DEGs) from microarray analysis of mRNA isolated from laser capture microdissection samples enriched for OB, lining cells, and osteocytes (OCy) revealed modules of genes that correlated with BFR/BS, BV/TV, and osteoblastic surface (Ob.S)/BS. Expression change of canonical Wnt target genes was similar in all three cell types at d8, including upregulation of Twist1 and Wisp1. At d29, the pattern of Wnt target gene expression changed in the OCy, with Twist1 returning to VEH level, sustained upregulation of Wisp1, and upregulation of several other Wnt targets that continued into the TFP. Predicted activation of pathways recognized to integrate with and regulate canonical Wnt signaling were also activated at d29 in the OCy. The most significantly affected pathways represented transcription factor signaling known to inhibit cell cycle progression (notably p53) and mitogenesis (notably c-Myc). These changes occurred at the time of peak BFR/BS and continued as BFR/BS declined during treatment, then trended toward VEH level in the TFP. Concurrent with this transcriptional switch was a reduction in OP numbers, an effect that would ultimately limit bone formation. This study confirms that the initial transcriptional response in response to Scl-Ab is activation of canonical Wnt signaling and the data demonstrate that there is induction of additional regulatory pathways in OCy with long-term treatment. The interactions between Wnt and p53/c-Myc signaling may be key in limiting OP populations, thus contributing to self-regulation of bone formation with continued Scl-Ab administration.

Sclerostin antibody (Scl-Ab) increases bone formation through a process dependent on the activation of canonical Wnt signaling, although the specific signaling in the osteoblast lineage in vivo is largely unknown. To gain insight into the signaling pathways acutely modulated by Scl-Ab, the transcriptional response of subpopulations of the osteoblast lineage was assessed by TaqMan and microarray analyses of mRNA isolated from laser capture microdissection (LCM)–enriched samples from the vertebrae of ovariectomized rats during the first week after Scl-Ab administration. Briefly, 6-month-old Sprague-Dawley rats were ovariectomized and, after 2 months, received a single dose of vehicle (VEH) or 100 mg/kg Scl-Ab (n = 20/group). Lumbar vertebrae were collected at 6, 24, 72, and 168 hours postdose and cryosectioned for LCM. Osteocytes were captured from bone matrix, and osteoblasts and lining cells were captured from bone surfaces based on fluorochrome labeling. mRNA was isolated, amplified, and profiled by TaqMan and microarray. Expression analysis revealed that Scl-Ab caused strikingly similar transcriptional profiles across all three cell types. Only 13 known canonical Wnt target genes, the majority with known functions in bone, showed a significant change in expression by microarray in response to Scl-Ab, with Wisp1 and Twist1 being the most responsive. Coincident with increased expression of Wnt target genes was the upregulation of numerous extracellular matrix (ECM) genes. The acute and progressive upregulation of ECM genes in lining cells supports their activation into matrix-producing osteoblasts, consistent with modeling-based bone formation. A similar transcriptional profile in osteocytes may indicate that Scl-Ab stimulates perilacunar/pericanalicular matrix deposition. Pathway analyses indicated that Scl-Ab regulated a limited number of genes related to cell cycle arrest and B-cell development. These data describe the acute downstream signaling in response to Scl-Ab in vivo and demonstrate selected canonical Wnt target gene activation associated with increased bone formation in all mature osteoblast subpopulations. © 2015 American Society for Bone and Mineral Research.

Mitochondrial perturbation has been recognized as a contributing factor to various drug-induced organ toxicities. To address this issue, we developed a high-throughput flow cytometry-based mitochondrial signaling assay to systematically investigate mitochondrial/cellular parameters known to be directly impacted by mitochondrial dysfunction: mitochondrial membrane potential (MMP), mitochondrial reactive oxygen species (ROS), intracellular reduced glutathione (GSH) level, and cell viability. Modulation of these parameters by a training set of compounds, comprised of established mitochondrial poisons and 60 marketed drugs (30 nM to 1mM), was tested in HL-60 cells (a human pro-myelocytic leukemia cell line) cultured in either glucose-supplemented (GSM) or glucose-free (containing galactose/glutamine; GFM) RPMI-1640 media. Post-hoc bio-informatic analyses of IC50 or EC50 values for all parameters tested revealed that MMP depolarization in HL-60 cells cultured in GSM was the most reliable parameter for determining mitochondrial dysfunction in these cells. Disruptors of mitochondrial function depolarized MMP at concentrations lower than those that caused loss of cell viability, especially in cells cultured in GSM; cellular GSH levels correlated more closely to loss of viability in vitro. Some mitochondrial respiratory chain inhibitors increased mitochondrial ROS generation; however, measuring an increase in ROS alone was not sufficient to identify mitochondrial disruptors. Furthermore, hierarchical cluster analysis of all measured parameters provided confirmation that MMP depletion, without loss of cell viability, was the key signature for identifying mitochondrial disruptors. Subsequent classification of compounds based on ratios of IC50s of cell viability:MMP determined that this parameter is the most critical indicator of mitochondrial health in cells and provides a powerful tool to predict whether novel small molecule entities possess this liability.

We previously reported an increased incidence of thrombotic toxicities in Sprague-Dawley rats administered the highest dose level of a hyperglycosylated analog of recombinant human erythropoietin (AMG 114) for 1 month as not solely dependent on high hematocrit (HCT). Thereafter, we identified increased erythropoiesis as a prothrombotic risk factor increased in the AMG 114 high-dose group with thrombotic toxicities, compared to a low-dose group with no toxicities but similar HCT. Here, we identified pleiotropic cytokines as prothrombotic factors associated with AMG 114 dose level. Before a high HCT was achieved, rats in the AMG 114 high, but not the low-dose group, had imbalanced hemostasis (increased von Willebrand factor and prothrombin time, decreased antithrombin III) coexistent with cytokines implicated in thrombosis: monocyte chemotactic protein 1 (MCP-1), MCP-3, tissue inhibitor of metalloproteinases 1, macrophage inhibitory protein-2, oncostatin M, T-cell-specific protein, stem cell factor, vascular endothelial growth factor, and interleukin-11. While no unique pathway to erythropoiesis stimulating agent-related thrombosis was identified, cytokines associated with increased erythropoiesis contributed to a prothrombotic intravascular environment in the AMG 114 high-dose group, but not in lower dose groups with a similar high HCT.

Muscle responses to exercise are complex and include acute responses to exercise-induced injury, as well as longer term adaptive training responses. Using Alaskan sled dogs as an experimental model, changes in muscle gene expression were analyzed to test the hypotheses that important regulatory elements of the muscle's adaptation to exercise could be identified based on the temporal pattern of gene expression. Dogs were randomly assigned to undertake a 160-km run ( n = 9), or to remain at rest ( n = 4). Biceps femoris muscle was obtained from the unexercised dogs and two dogs at each of 2, 6, and 12 h after the exercise, and from three dogs 24 h after exercise. RNA was extracted and microarray analysis used to define gene transcriptional changes. The changes in gene expression after exercise occurred in a temporal pattern. Overall, 569, 469, 316, and 223 transcripts were differentially expressed at 2, 6, 12, and 24 h postexercise, respectively, compared with unexercised dogs (based on P ≤ 0.01 and an absolute fold change of ≥1.5). Increases in a number of known transcriptional regulators, including peroxisome proliferator-activated receptor-α, cAMP-responsive element modulator, and CCAAT enhancer binding protein-δ, and potential signaling molecules, including brain-derived neurotrophic factor, dermokine, and suprabasin, were observed 2 h after exercise. Biological functional analysis suggested changes in expression of genes with known functional relationships, including genes involved in muscle remodeling and growth, intermediary metabolism, and immune regulation. Sustained endurance exercise by Alaskan sled dogs induces coordinated changes in gene expression with a clear temporal pattern. RNA expression profiling has the potential to identify novel regulatory mechanisms and responses to exercise stimuli.

Oligonucleotide microarrays were used to study the variability of pharmacokinetics and drug metabolism (PKDM)-related gene expression in 75 normal human livers. The objective was to define and use absorption, distribution, metabolism and excretion (ADME) gene expression variability to discern co-regulated genes and potential surrogate biomarkers of inducible gene expression. RNA was prepared from donor tissue and hybridized on Agilent microarrays against an RNA mass balanced pool from all donors. Clustering of PKDM gene sets revealed donors with distinct patterns of gene expression that grouped genes known to be regulated by the nuclear receptor, pregnane X-receptor (PXR). Fold range metrics and frequency distributions from the heterogeneous human population were used to define the variability of individual PKDM genes in the 75 human livers and were placed in context by comparing expression data with basal ADME gene expression variability in an inbred and diet/environment controlled population of 27 Rhesus livers. The most variable genes in the hepatic transcriptome were mainly related to drug metabolism, intermediary metabolism, inflammation and cell cycle control. Unique patterns of expression across 75 individuals of inducible ADME gene expression allowed their expression to be correlated with the expression of many other genes. Correlated genes for AhR, CAR and PXR responsive genes (CYP1A2, CYP2B6 and CYP3A4) were identified that may be co-regulated and, therefore, provide clues to the identity of surrogate gene or protein markers for CYP induction. In conclusion, microarrays were used to define the variable expression of hepatic ADME genes in a diverse human population, the expression variability of ADME genes was compared with the expression variability in an inbred population of Rhesus monkeys, and genes were defined that may be co-regulated with important inducible CYP genes.

Motivation: Microarray gene expression data are increasingly employed to identify sets of marker genes that accurately predict disease development and outcome in cancer. Many computational approaches have been proposed to construct such predictors. However, there is, as yet, no objective way to evaluate whether a new approach truly improves on the current state of the art. In addition no 'standard' computational approach has emerged which enables robust outcome prediction. Results: An important contribution of this work is the description of a principled training and validation protocol, which allows objective evaluation of the complete methodology for constructing a predictor. We review the possible choices of computational approaches, with specific emphasis on predictor choice and reporter selection strategies. Employing this training-validation protocol, we evaluated different reporter selection strategies and predictors on six gene expression datasets of varying degrees of difficulty. We demonstrate that simple reporter selection strategies (forward filtering and shrunken centroids) work surprisingly well and outperform partial least squares in four of the six datasets. Similarly, simple predictors, such as the nearest mean classifier, outperform more complex classifiers. Our training-validation protocol provides a robust methodology to evaluate the performance of new computational approaches and to objectively compare outcome predictions on different datasets. Contact: l.f.a.wessels@ewi.tudelft.nl Supplementary information: http://ict.ewi.tudelft.nl/index.php?option=com_pub&task=view&id=1983

Based on the hypothesis that features of the molecular program of normal wound healing might play an important role in cancer metastasis, we previously identified consistent features in the transcriptional response of normal fibroblasts to serum, and used this “wound-response signature” to reveal links between wound healing and cancer progression in a variety of common epithelial tumors. Here, in a consecutive series of 295 early breast cancer patients, we show that both overall survival and distant metastasis-free survival are markedly diminished in patients whose tumors expressed this wound-response signature compared to tumors that did not express this signature. A gene expression centroid of the wound-response signature provides a basis for prospectively assigning a prognostic score that can be scaled to suit different clinical purposes. The wound-response signature improves risk stratification independently of known clinico-pathologic risk factors and previously established prognostic signatures based on unsupervised hierarchical clustering (“molecular subtypes”) or supervised predictors of metastasis (“70-gene prognosis signature”).

Ligand-mediated activation of the pregnane X receptor (PXR, NR1I2) is postulated to affect both hepatic and intestinal gene expression, because of the presence of this nuclear receptor in these important drug metabolizing organs; as such, activation of this receptor may elicit the coordinated regulation of PXR target genes in both tissues. Induction of hepatic and intestinal drug metabolism can contribute to the increased metabolism of drugs, and can result in adverse or undesirable drug-drug interactions. 2(S)-((3,5-bis(Trifluoromethyl)benzyl)-oxy)-3(S)phenyl-4-((3-oxo-1,2,4-triazol-5-yl)methyl)morpholine (L-742694) is a potent activator of the rat PXR and was characterized for its effects on hepatic and intestinal gene expression in female Sprague-Dawley rats by DNA microarray analysis. Transcriptional profiling in liver and small intestine revealed that L-742694 and dexamethasone (DEX) induced the prototypical battery of PXR target genes in liver, including CYP3A, Oatp2, and UGT1A1. In addition, both DEX and L-742694 induced common gene expression profiles that were specific to liver or small intestine, but there was a distinct lack of coordinated gene expression of genes common to both tissues. This pattern of gene regulation occurred in liver and small intestine independent of PXR, constitutive androstane receptor, or hepatic nuclear factor-4alpha expression, suggesting that other factors are involved in controlling the extent of coordinated gene expression in response to a PXR agonist. Overall, these results suggest that ligand-mediated activation of PXR and induction of hepatic, rather than small intestinal, drug metabolism genes would contribute to the increased metabolism of orally administered pharmaceuticals.

A data anomaly was observed that affected the uniformity and reproducibility of fluorescent signal across DNA microarrays. Results from experimental sets designed to identify potential causes (from microarray production to array scanning) indicated that the anomaly was linked to a batch process; further work allowed us to localize the effect to the posthybridization array stringency washes. Ozone levels were monitored and highly correlated with the batch effect. Controlled exposures of microarrays to ozone confirmed this factor as the root cause, and we present data that show susceptibility of a class of cyanine dyes (e.g., Cy5, Alexa 647) to ozone levels as low as 5−10 ppb for periods as short as 10−30 s. Other cyanine dyes (e.g., Cy3, Alexa 555) were not significantly affected until higher ozone levels (>100 ppb). To address this environmental effect, laboratory ozone levels should be kept below 2 ppb (e.g., with filters in HVAC) to achieve high quality microarray data.

Motivation: There is a very large and growing level of effort toward improving the platforms, experiment designs, and data analysis methods for microarray expression profiling. Along with a growing richness in the approaches there is a growing confusion among most scientists as to how to make objective comparisons and choices between them for different applications. There is a need for a standard framework for the microarray community to compare and improve analytical and statistical methods. Results: We report on a microarray data set comprising 204 in-situ synthesized oligonucleotide arrays, each hybridized with two-color cDNA samples derived from 20 different human tissues and cell lines. Design of the ∼ 24 000 60mer oligonucleotides that report ∼2500 known genes on the arrays, and design of the hybridization experiments, were carried out in a way that supports the performance assessment of alternative data processing approaches and of alternative experiment and array designs. We also propose standard figures of merit for success in detecting individual differential expression changes or expression levels, and for detecting similarities and differences in expression patterns across genes and experiments. We expect this data set and the proposed figures of merit will provide a standard framework for much of the microarray community to compare and improve many analytical and statistical methods relevant to microarray data analysis, including image processing, normalization, error modeling, combining of multiple reporters per gene, use of replicate experiments, and sample referencing schemes in measurements based on expression change. Availability/Supplementary information: Expression data and supplementary information are available at http://www.rii.com/publications/2003/HE_SDS.htm Contact: yudong_he@merck.com * To whom correspondence should be addressed. † A wholly owned subsidiary of Merck & Co., Inc.

Background A more accurate means of prognostication in breast cancer will improve the selection of patients for adjuvant systemic therapy. Methods Using microarray analysis to evaluate our previously established 70-gene prognosis profile, we classified a series of 295 consecutive patients with primary breast carcinomas as having a gene-expression signature associated with either a poor prognosis or a good prognosis. All patients had stage I or II breast cancer and were younger than 53 years old; 151 had lymph-node–negative disease, and 144 had lymph-node–positive disease. We evaluated the predictive power of the prognosis profile using univariable and multivariable statistical analyses. Results Among the 295 patients, 180 had a poor-prognosis signature and 115 had a good-prognosis signature, and the mean (±SE) overall 10-year survival rates were 54.6±4.4 percent and 94.5±2.6 percent, respectively. At 10 years, the probability of remaining free of distant metastases was 50.6±4.5 percent in the group with a poor-prognosis signature and 85.2±4.3 percent in the group with a good-prognosis signature. The estimated hazard ratio for distant metastases in the group with a poor-prognosis signature, as compared with the group with the good-prognosis signature, was 5.1 (95 percent confidence interval, 2.9 to 9.0; P

We describe a flexible system for gene expression profiling using arrays of tens of thousands of oligonucleotides synthesized in situ by an ink-jet printing method employing standard phosphoramidite chemistry. We have characterized the dependence of hybridization specificity and sensitivity on parameters including oligonucleotide length, hybridization stringency, sequence identity, sample abundance, and sample preparation method. We find that 60-mer oligonucleotides reliably detect transcript ratios at one copy per cell in complex biological samples, and that ink-jet arrays are compatible with several different sample amplification and labeling techniques. Furthermore, results using only a single carefully selected oligonucleotide per gene correlate closely with those obtained using complementary DNA (cDNA) arrays. Most of the genes for which measurements differ are members of gene families that can only be distinguished by oligonucleotides. Because different oligonucleotide sequences can be specified for each array, we anticipate that ink-jet oligonucleotide array technology will be useful in a wide variety of DNA microarray applications.

Genome-wide transcript profiling was used to monitor signal transduction during yeast pheromone response. Genetic manipulations allowed analysis of changes in gene expression underlying pheromone signaling, cell cycle control, and polarized morphogenesis. A two-dimensional hierarchical clustered matrix, covering 383 of the most highly regulated genes, was constructed from 46 diverse experimental conditions. Diagnostic subsets of coexpressed genes reflected signaling activity, cross talk, and overlap of multiple mitogen-activated protein kinase (MAPK) pathways. Analysis of the profiles specified by two different MAPKs—Fus3p and Kss1p—revealed functional overlap of the filamentous growth and mating responses. Global transcript analysis reflects biological responses associated with the activation and perturbation of signal transduction pathways.

We recently reported results that erythropoiesis-stimulating agent (ESA)–related thrombotic toxicities in preclinical species were not solely dependent on a high hematocrit (HCT) but also associated with increased ESA dose level, dose frequency, and dosing duration. In this article, we conclude that sequelae of an increased magnitude of ESA-stimulated erythropoiesis potentially contributed to thrombosis in the highest ESA dose groups. The results were obtained from two investigative studies we conducted in Sprague-Dawley rats administered a low (no thrombotic toxicities) or high (with thrombotic toxicities) dose level of a hyperglycosylated analog of recombinant human erythropoietin (AMG 114), 3 times weekly for up to 9 days or for 1 month. Despite similarly increased HCT at both dose levels, animals in the high-dose group had an increased magnitude of erythropoiesis measured by spleen weights, splenic erythropoiesis, and circulating reticulocytes. Resulting prothrombotic risk factors identified predominantly or uniquely in the high-dose group were higher numbers of immature reticulocytes and nucleated red blood cells in circulation, severe functional iron deficiency, and increased intravascular destruction of iron-deficient reticulocyte/red blood cells. No thrombotic events were detected in rats dosed up to 9 days suggesting a sustained high HCT is a requisite cofactor for development of ESA-related thrombotic toxicities.

A common inflammatome signature, as well as disease-specific expression patterns, was identified from 11 different rodent inflammatory disease models. Causal regulatory networks and the drivers of the inflammatome signature were uncovered and validated., Representative inflammatome gene signatures, as well as disease model-specific gene signatures, were identified from 12 gene expression profiling data sets derived from 9 different tissues isolated from 11 rodent inflammatory disease models. The inflammatome signature is highly enriched for immune response-related genes, disease causal genes, and drug targets. Regulatory relationships among the inflammatome signature genes were examined in over 70 causal networks derived from a number of large-scale genetic studies of multiple diseases, and the potential key drivers were uncovered and validated prospectively. Over 70% of the inflammatome signature genes and over 50% of the key driver genes have not been reported in previous studies of common signatures in inflammatory conditions., Common inflammatome gene signatures as well as disease-specific signatures were identified by analyzing 12 expression profiling data sets derived from 9 different tissues isolated from 11 rodent inflammatory disease models. The inflammatome signature significantly overlaps with known drug targets and co-expressed gene modules linked to metabolic disorders and cancer. A large proportion of genes in this signature are tightly connected in tissue-specific Bayesian networks (BNs) built from multiple independent mouse and human cohorts. Both the inflammatome signature and the corresponding consensus BNs are highly enriched for immune response-related genes supported as causal for adiposity, adipokine, diabetes, aortic lesion, bone, muscle, and cholesterol traits, suggesting the causal nature of the inflammatome for a variety of diseases. Integration of this inflammatome signature with the BNs uncovered 151 key drivers that appeared to be more biologically important than the non-drivers in terms of their impact on disease phenotypes. The identification of this inflammatome signature, its network architecture, and key drivers not only highlights the shared etiology but also pinpoints potential targets for intervention of various common diseases.

Genome-wide gene expression profiling has become standard for assessing potential liabilities as well as for elucidating mechanisms of toxicity of drug candidates under development. Analysis of microarray data is often challenging due to the lack of a statistical model that is amenable to biological variation in a small number of samples. Here we present a novel non-parametric algorithm that requires minimal assumptions about the data distribution. Our method for determining differential expression consists of two steps: 1) We apply a nominal threshold on fold change and platform p-value to designate whether a gene is differentially expressed in each treated and control sample relative to the averaged control pool, and 2) We compared the number of samples satisfying criteria in step 1 between the treated and control groups to estimate the statistical significance based on a null distribution established by sample permutations. The method captures group effect without being too sensitive to anomalies as it allows tolerance for potential non-responders in the treatment group and outliers in the control group. Performance and results of this method were compared with the Significant Analysis of Microarrays (SAM) method. These two methods were applied to investigate hepatic transcriptional responses of wild-type (PXR(+/+)) and pregnane X receptor-knockout (PXR(-/-)) mice after 96 h exposure to CMP013, an inhibitor of β-secretase (β-site of amyloid precursor protein cleaving enzyme 1 or BACE1). Our results showed that CMP013 led to transcriptional changes in hallmark PXR-regulated genes and induced a cascade of gene expression changes that explained the hepatomegaly observed only in PXR(+/+) animals. Comparison of concordant expression changes between PXR(+/+) and PXR(-/-) mice also suggested a PXR-independent association between CMP013 and perturbations to cellular stress, lipid metabolism, and biliary transport.

Toxicity is a major cause of failure in drug development. A toxicogenomic approach may provide a powerful tool for better assessing the potential toxicity of drug candidates. Several approaches have been reported for predicting hepatotoxicity based on reference compounds with well-studied toxicity mechanisms. We developed a new approach for assessing compound-induced liver injury without prior knowledge of a compound's mechanism of toxicity. Using samples from rodents treated with 49 known liver toxins and 10 compounds without known liver toxicity, we derived a hepatotoxicity score as a single quantitative measurement for assessing the degree of induced liver damage. Combining the sensitivity of the hepatotoxicity score and the power of a machine learning algorithm, we then built a model to predict compound-induced liver injury based on 212 expression profiles. As estimated in an independent data set of 54 expression profiles, the built model predicted compound-induced liver damage with 90.9% sensitivity and 88.4% specificity. Our findings illustrate the feasibility of ab initio estimation of liver toxicity based on transcriptional profiles.

The molecular mechanisms of action of a HIV protease inhibitor, ritonavir, on hepatic function were explored on a genomic scale using microarrays comprising genes expressed in the liver of Sprague–Dawley rats (Rattus norvegicus). Analyses of hepatic transcriptional fingerprints led to the identification of several key cellular pathways affected by ritonavir treatment. These effects were compared to a compendium of gene expression responses for 52 unrelated compounds and to other protease inhibitors, including atazanavir and two experimental compounds. We identified genes involved in cholesterol and fatty acid biosynthesis, as well as genes involved in fatty acid and cholesterol breakdown, whose expressions were regulated in opposite manners by ritonavir and bezafibrate, a hypolipidemic agonist of the peroxisome proliferator-activated receptor α. Ritonavir also upregulated multiple proteasomal subunit transcripts as well as genes involved in ubiquitination, consistent with its known inhibitory effect on proteasomal activity. We also tested three other protease inhibitors in addition to ritonavir. Atazanavir did not impact ubiquitin or proteasomal gene expression, although the two other experimental protease inhibitors impacted both proteasomal gene expression and sterol regulatory element-binding protein-activated genes, similar to ritonavir. Identification of key metabolic pathways that are affected by ritonavir and other protease inhibitors will enable us to understand better the downstream effects of protease inhibitors, thus leading to better drug design and an effective method to mitigate the side effects of this important class of HIV therapeutics.

Detecting and understanding the potential for off-target pharmacological effects is critical in the optimization of lead compounds in drug discovery programs. Compound-mediated activation of the pregnane X receptor (PXR; NR1I2), a key regulator for drug metabolism genes, is often monitored to avoid potential drug-drug interactions. Two structural analogs, MRL-1 and MRL-2, were determined to be equivalent PXR activators in trans-activation assays. To differentiate these two PXR activators, their transcriptional effects were examined in PXR-sufficient (LS180) and PXR-deficient (Caco-2) adenocarcinoma cell lines. Both compounds regulated drug-management genes (e.g. CYP3A4, CYP2B6, UGT1A1 and ABCB1) in LS180 cells, but not in PXR-deficient Caco-2 cells. The potency of MRL-1 and MRL-2 on PXR activation was again equivalent as revealed by a set of 113 genes that were regulated by four prototypical PXR agonists (rifampicin, ritonavir, troglitazone and dexamethasone) in the LS180 cells. The specificity of the PXR signature genes was supported by the enrichment of putative PXR binding sites uncovered by sequence-based promoter analyses. Interestingly, an additional off-target activity of MRL-2 was suggested where sterol response element binding protein binding sites were found enriched in a subset of PXR signature genes. These genes, involved in cholesterol and fatty acid synthesis, were significantly regulated by ritonavir, chlorpromazine and MRL-2, which were linked to the manifestation of phospholipidosis. The present study demonstrates the utility of our approach in the differentiation and selection of lead compounds for drug development.

The antiandrogen flutamide (FLU) is used primarily for prostate cancer and is an idiosyncratic hepatotoxicant that sometimes causes severe liver problems. To investigate FLU's overt hepatic effects, especially on inducible drug clearance-related gene networks, FLU's hepatic gene expression profile was examined in female Sprague-Dawley rats using approximately 22,500 oligonucleotide microarrays. Rats were dosed daily for 3 days with FLU at 500, 250, 62.5, 31.3, and 15.6 mg/kg/day, and hepatic RNA was isolated. FLU resulted in the dose-dependent regulation of approximately 350 genes. Employing a gene-response compendium, FLU was compared with three classical aryl hydrocarbon receptor (AhR) ligands, 3-methylcholanthrene, benzo[a]pyrene, and beta-naphthoflavone, and four atypical CYP1A inducers, indole-3-carbinol (I3C), omeprazole (OME), chlorpromazine (CPZ), and clotrimazole (CLO). The FLU gene response was comparable with classical AhR ligands across a signature AhR ligand gene set that included CYP1A1 and other members of the AhR gene battery. Dose-related responses of CYP1 genes established a maximum response ceiling and discerned potency differences in atypical inducers. FLU had a sharp down-regulation of c-fos that was comparable with all the compounds except CPZ and CLO. FLU absorption, distribution, metabolism, and excretion (ADME) gene expression analysis revealed that FLU, as well as I3C and OME, induced CYP2B and CYP3A, distinguishing them from the classical AhR ligands. By using a compendium of gene expression profiles, FLU was shown to signal in rats similar to an AhR activator with additional CYP2B and CYP3A effects that most resembled the ADME gene expression pattern of the atypical CYP1A inducers I3C and OME.

High-capacity methods for assessing gene function have become increasingly important because of the increasing number of newly identified genes emerging from large-scale genome sequencing and cDNA cloning efforts. We investigated the use of DNA microarrays to identify uncharacterized genes specifically involved in human T cell activation. Activation of human peripheral blood T lymphocytes induced significant changes in hundreds of transcripts, but most of these were not unique to T cell activation. Variation of experimental parameters and analysis techniques allowed better enrichment for gene expression changes unique to T cell activation. Best results were achieved by identification of genes that were most highly coregulated with the T-cell-specific transcript interleukin 2 (IL2) in a “compendium” of experiments involving both T cells and other cell types. Among the genes most highly coregulated with IL2 were many genes known to function during T cell activation, together with ESTs of unknown function. Four of these ESTs were extended to novel full-length clones encoding T-cell-regulated proteins with predicted functions in GTP metabolism, cell organization, and signal transduction.

Gruvberger et al. postulate, in their commentary [1] published in this issue of Breast Cancer Research, that our "prognostic gene set may not be broadly applicable to other breast tumor cohorts", and they suggest that "it may be important to define prognostic expression profiles separately in estrogen receptor (ER) positive and negative tumors". This is based on two observations derived from our gene expression profiling data in breast cancer [2]: the overlap between reporter genes for prognosis and ER status, and Gruvberger et al.'s inability to confirm the prognosis prediction using a nonoptimal selection of 58 of our 231 prognosis reporter genes. The overlap between our prognosis reporter genes and the ER status genes is certainly very large, mainly because ~10% of all genes on our microarray contain information on the ER status (2460 out of 24,479). However, the overlap between the 70 optimal prognosis genes and the 550 optimal ER status genes is only 17% (12 out of 70). We therefore believe that there is a different subset of genes that reports prognosis as compared with ER status. Our prognosis classifier strongly predicts the risk of distant metastases (odds ratio = 15, 95% confidence interval = 4–56, P < 0.0001). Adjusted for associations with known clinical prognosticators, including ER status, in a multivariable analysis, this odds ratio slightly increases to 18 (95% confidence interval = 3.3–94, P = 0.0002) [2]. This indicates that the predictive capacity of our prognosis classifier cannot be explained by its association with, among other factors, ER status as suggested. There are a few points of concern relating to Gruvberger et al.'s inability to develop an outcome classifier on their data set. It should be pointed out that the prognosis signature is subtle both in the number of genes and the magnitude of differential regulation as compared with the ER status signature. The microarray platform sensitivity and reproducibility is a key issue in generating high quality data that ultimately determine the power of discovery. The reproducibility of our platform is clearly demonstrated by the uniform patterns related to ER status (Fig. 3b in [2]) as compared with the ER status expression patterns published by Gruvberger et al. (Fig. 2b in [3]). Our experiences with cDNA arrays and oligonucleotide arrays showed that the cDNA arrays are less sensitive in detecting small differential regulations because of the high chance of nonspecific binding. Another point concerns the reference sample used in the two-color assay and its effect on the sensitivity. A nearby reference (a tumor sample pool in our case) makes the differences between the samples easily detectable, whereas a distant reference (one cell line [1]) makes this harder to detect since the fluctuations of the large offset interferes with the measurement of small differential signals. Moreover, we always repeat the same measurement twice in fluor-reversal pairs to minimise the potential labeling biases. It is possible that the aforementioned reasons and the fact that most of our 70 optimal genes were not on their array prevented Gruvberger et al. from revealing the prognosis signature. Finally, it is of importance to note that we derived our prognosis classifier from breast cancer patients of whom the majority (93%) did not receive adjuvant systemic therapy, whereas all Gruvberger et al.'s patients received adjuvant tamoxifen treatment. So, a relatively better outcome (approximately 30%) can be expected within Gruvberger et al.'s ER-positive subgroup because of the tamoxifen treatment. The predictive power of our prognosis reporters may be reduced in an adjuvantly treated patient group. A confirmation on a large unselected 'cohort' of breast cancer patients is required to validate our findings. We have recently completed DNA microarray analyses on a cohort of 295 breast cancer samples. The results of this will be published in the near future.

Gene expression profiling is a tool to gain mechanistic understanding of adverse effects in response to compound exposure. However, little is known about how the common handling procedures of experimental animals during a preclinical study alter baseline gene expression. We report gene expression changes in the livers of female Sprague-Dawley rats following common handling procedures. Baseline gene expression changes identified in this study provide insight on how these changes may affect interpretation of gene expression profiles following compound exposure. Rats were divided into three groups. One group was not subjected to handling procedures and served as controls for both handled groups. Animals in the other two groups were weighed, subjected to restraint in Broome restrainers, and administered water via oral gavage daily for 1 or 4 days with tail vein blood collections at 1, 2, 4, and 8 hours postdose on days 1 and 4. Significantly altered genes were identified in livers of animals following 1 or 4 days of handling when compared to the unhandled animals. Gene changes in animals handled for 4 days were similar to those handled for 1 day, suggesting a lack of habituation. The altered genes were primarily immune function related genes. These findings, along with a correlating increase in corticosterone levels suggest that common handling procedures may cause a minor immune system perturbance.

The most important product of the sequencing of a genome is a complete, accurate catalogue of genes and their products, primarily messenger RNA transcripts and their cognate proteins. Such a catalogue cannot be constructed by computational annotation alone; it requires experimental validation on a genome scale. Using ‘exon’ and ‘tiling’ arrays fabricated by ink-jet oligonucleotide synthesis, we devised an experimental approach to validate and refine computational gene predictions and define full-length transcripts on the basis of co-regulated expression of their exons. These methods can provide more accurate gene numbers and allow the detection of mRNA splice variants and identification of the tissue- and disease-specific conditions under which genes are expressed. We apply our technique to chromosome 22q under 69 experimental condition pairs, and to the entire human genome under two experimental conditions. We discuss implications for more comprehensive, consistent and reliable genome annotation, more efficient, full-length complementary DNA cloning strategies and application to complex diseases.

Ascertaining the impact of uncharacterized perturbations on the cell is a fundamental problem in biology. Here, we describe how a single assay can be used to monitor hundreds of different cellular functions simultaneously. We constructed a reference database or “compendium” of expression profiles corresponding to 300 diverse mutations and chemical treatments in S. cerevisiae, and we show that the cellular pathways affected can be determined by pattern matching, even among very subtle profiles. The utility of this approach is validated by examining profiles caused by deletions of uncharacterized genes: we identify and experimentally confirm that eight uncharacterized open reading frames encode proteins required for sterol metabolism, cell wall function, mitochondrial respiration, or protein synthesis. We also show that the compendium can be used to characterize pharmacological perturbations by identifying a novel target of the commonly used drug dyclonine.