Search

Supramolecular signaling assemblies are of interest for their unique signaling properties. A µm scale signaling assembly, the central supramolecular signaling cluster (cSMAC), forms at the center of the interface of T cells activated by antigen-presenting cells. We have determined that it is composed of multiple complexes of a supramolecular volume of up to 0.5 µm3 and associated with extensive membrane undulations. To determine cSMAC function, we have systematically manipulated the localization of three adaptor proteins, LAT, SLP-76, and Grb2. cSMAC localization varied between the adaptors and was diminished upon blockade of the costimulatory receptor CD28 and deficiency of the signal amplifying kinase Itk. Reconstitution of cSMAC localization restored IL-2 secretion which is a key T cell effector function as dependent on reconstitution dynamics. Our data suggest that the cSMAC enhances early signaling by facilitating signaling interactions and attenuates signaling thereafter through sequestration of a more limited set of signaling intermediates., eLife digest Cells receive dozens of signals at different times and in different places. Integrating incoming information and deciding how to respond is no easy task. Signaling molecules on the cell surface pass messages inwards using chemical messengers that interact in complicated networks within the cell. One way to unravel the complexity of these networks is to look at specific groups of signaling molecules in test tubes to see how they interact. But the interior of a living cell is a very different environment. Molecules inside cells are tightly packed and, under certain conditions, they interact with each other by the thousands. They form structures known as ‘supramolecular complexes’, which changes their behavior. One such supramolecular complex is the ‘central supramolecular activation cluster’, or cSMAC for short. It forms under the surface of immune cells called T cells when they are getting ready to fight an infection. Under the microscope, the cSMAC looks like the bullseye of a dartboard, forming a crowd of signaling molecules at the center of the interface between the T cell and another cell. Its exact role is not clear, but evidence suggests it helps to start and stop the signals that switch T cells on. The cSMAC contains two key protein adaptors called LAT and SLP-76 that help to hold the structure together. So, to find out what the cSMAC does, Clark et al. genetically modified these adaptors to gain control over when the cSMAC forms. Clark et al. examined mouse T cells using super-resolution microscopy and electron microscopy, watching as other immune cells delivered the signal to switch on. As the T cells started to activate, the composition of the cSMAC changed. In the first two minutes after the cells started activating, the cSMAC included a large number of different components. This made T cell activation more efficient, possibly because the supramolecular complex was helping the network of signals to interact. Later, the cSMAC started to lose many of these components. Separating components may have helped to stop the activation signals. Understanding how T cells activate could lead to the possibility of turning them on or off in immune-related diseases. But these findings are not just relevant to immune cells. Other cells also use supramolecular complexes to control their signaling. Investigating how these complexes change over time could help us to understand how other cell types make decisions.

Protein complexes play a significant role in the core functionality of cells. These complexes are typically identified by detecting densely connected subgraphs in protein-protein interaction (PPI) networks. Recently, multiple large-scale mass spectrometry-based experiments have significantly increased the availability of PPI data in order to further expand the set of known complexes. However, high-throughput experimental data generally are incomplete, show limited agreement between experiments, and show frequent false positive interactions. There is a need for computational approaches that can address these limitations in order to improve the coverage and accuracy of human protein complexes. Here, we present a new method that integrates data from multiple heterogeneous experiments and sources in order to increase the reliability and coverage of predicted protein complexes. We first fused the heterogeneous data into a feature matrix and trained classifiers to score pairwise protein interactions. We next used graph based methods to combine pairwise interactions into predicted protein complexes. Our approach improves the accuracy and coverage of protein pairwise interactions, accurately identifies known complexes, and suggests both novel additions to known complexes and entirely new complexes. Our results suggest that integration of heterogeneous experimental data helps improve the reliability and coverage of diverse high-throughput mass-spectrometry experiments, leading to an improved global map of human protein complexes.

PURPOSE OF REVIEW: Spinal deformity is a common issue in pediatric patients with an underlying neurological diagnosis or syndrome. Management of neuromuscular scoliosis (NMS) is a major part of the orthopedic care of such patients, as the deformity is often progressive, and may affect gait, seating and positioning. In addition, untreated large spinal deformities may be associated with pain and/or cardiopulmonary issues over time. RECENT FINDINGS: Recent changes in medical management of the underlying disease process appears to alter the natural history of certain neuromuscular conditions, and in the case of patients with Duchenne’s muscular dystrophy significantly diminish the incidence of spinal deformity. In the most common diagnosis associated with NMS, cerebral palsy, there is evidence that despite a high complication rate, surgical management of spinal deformity is associated with measurable improvements in validated health-related quality-of-life measures. SUMMARY: Spinal deformity is a common finding in patients with neurological diagnoses. It is important for those involved in the care of these patients to understand the natural history of NMS, as well as the potential risks and benefits to the patient and caregivers, of surgical and non-surgical interventions.

Supramolecular signaling assemblies are of interest for their unique signaling properties. A µm scale signaling assembly, the central supramolecular signaling cluster (cSMAC), forms at the center of the interface of T cells activated by antigen presenting cells. We have determined that it is composed of multiple complexes of a supramolecular volume of up to 0.5µm3 and associated with extensive membrane undulations. To determine cSMAC function, we have systematically manipulated the localization of three adaptor proteins, LAT, SLP-76, and Grb2. cSMAC localization varied between the adaptors and was diminished upon blockade of the costimulatory receptor CD28 and deficiency of the signal amplifying kinase Itk. Reconstitution of cSMAC localization restored IL-2 secretion which is a key T cell effector function as dependent on reconstitution dynamics. Our data suggest that the cSMAC enhances early signaling by facilitating signaling interactions and attenuates signaling thereafter through sequestration of a more limited set of signaling intermediates.

It is shown for the first time that cell shape can be accurately predicted from nuclear shape (and vice versa) for three different cell lines. This correlation is reduced by altering protein C1QBP or various drugs. In addition, a generative model is given for the kinetics of shape change. The software is available in the open-source CellOrganizer system., Modeling cell shape variation is critical to our understanding of cell biology. Previous work has demonstrated the utility of nonrigid image registration methods for the construction of nonparametric nuclear shape models in which pairwise deformation distances are measured between all shapes and are embedded into a low-dimensional shape space. Using these methods, we explore the relationship between cell shape and nuclear shape. We find that these are frequently dependent on each other and use this as the motivation for the development of combined cell and nuclear shape space models, extending nonparametric cell representations to multiple-component three-dimensional cellular shapes and identifying modes of joint shape variation. We learn a first-order dynamics model to predict cell and nuclear shapes, given shapes at a previous time point. We use this to determine the effects of endogenous protein tags or drugs on the shape dynamics of cell lines and show that tagged C1QBP reduces the correlation between cell and nuclear shape. To reduce the computational cost of learning these models, we demonstrate the ability to reconstruct shape spaces using a fraction of computed pairwise distances. The open-source tools provide a powerful basis for future studies of the molecular basis of cell organization.

Macroautophagy is regarded as a nonspecific bulk degradation process of cytoplasmic material within the lysosome. However, the process has mainly been studied by nonspecific bulk degradation assays using radiolabeling. In the present study we monitor protein turnover and degradation by global, unbiased approaches relying on quantitative mass spectrometry-based proteomics. Macroautophagy is induced by rapamycin treatment, and by amino acid and glucose starvation in differentially, metabolically labeled cells. Protein dynamics are linked to image-based models of autophagosome turnover. Depending on the inducing stimulus, protein as well as organelle turnover differ. Amino acid starvation-induced macroautophagy leads to selective degradation of proteins important for protein translation. Thus, protein dynamics reflect cellular conditions in the respective treatment indicating stimulus-specific pathways in stress-induced macroautophagy.

Imaging techniques such as immunofluorescence (IF) and the expression of fluorescent protein (FP) fusions are widely used to investigate the subcellular distribution of proteins. Here we report a systematic analysis of >500 human proteins comparing the localizations obtained in live versus fixed cells using FPs and IF, respectively. We identify systematic discrepancies between IF and FPs as well as between FP tagging at the N and C termini. The analysis shows that for 80% of the proteins, IF and FPs yield the same subcellular distribution, and the locations of 250 previously unlocalized proteins were determined by the overlap between the two methods. Approximately 60% of proteins localize to multiple organelles for both methods, indicating a complex subcellular protein organization. These results show that both IF and FP tagging are reliable techniques and demonstrate the usefulness of an integrative approach for a complete investigation of the subcellular human proteome. QC 20130521

Fluorescence microscopy is one of the most important tools in cell biology research because it provides spatial and temporal information to investigate regulatory systems inside cells. This technique can generate data in the form of signal intensities at thousands of positions resolved inside individual live cells. However, given extensive cell-to-cell variation, these data cannot be readily assembled into three- or four-dimensional maps of protein concentration that can be compared across different cells and conditions. We have developed a method to enable comparison of imaging data from many cells and applied it to investigate actin dynamics in T cell activation. Antigen recognition in T cells by the T cell receptor (TCR) is amplified by engagement of the costimulatory receptor CD28. We imaged actin and eight core actin regulators to generate over a thousand movies of T cells under conditions in which CD28 was either engaged or blocked in the context of a strong TCR signal. Our computational analysis showed that the primary effect of costimulation blockade was to decrease recruitment of the activator of actin nucleation WAVE2 (Wiskott-Aldrich syndrome protein family verprolin-homologous protein 2) and the actin-severing protein cofilin to F-actin. Reconstitution of WAVE2 and cofilin activity restored the defect in actin signaling dynamics caused by costimulation blockade. Thus, we have developed and validated an approach to quantify protein distributions in time and space for the analysis of complex regulatory systems.

A recent meeting entitled Frontiers in Live Cell Imaging was attended by more than 400 cell biologists, physicists, chemists, mathematicians, and engineers. Unlike typical special topics meetings, which bring together investigators in a defined field primarily to review recent progress, the purpose of this meeting was to promote cross-disciplinary interactions by introducing emerging methods on the one hand and important biological applications on the other. The goal was to turn live cell imaging from a “technique” used in cell biology into a new exploratory science that combines a number of research fields.

The goal of proteomics is the complete characterization of all proteins. Efforts to characterize subcellular location have been limited to assigning proteins to general categories of organelles. We have previously designed numerical features to describe location patterns in microscope images and developed automated classifiers that distinguish major subcellular patterns with high accuracy (including patterns not distinguishable by visual examination). The results suggest the feasibility of automatically determining which proteins share a single location pattern in a given cell type. We describe an automated method that selects the best feature set to describe images for a given collection of proteins and constructs an effective partitioning of the proteins by location. An example for a limited protein set is presented. As additional data become available, this approach can produce for the first time an objective systematics for protein location and provide an important starting point for discovering sequence motifs that determine localization.

Detection of neuronal cell differentiation is essential to study cell fate decisions under various stimuli and/or environmental conditions. Many tools exist that quantify differentiation by neurite length measurements of single cells. However, quantification of differentiation in whole cell populations remains elusive so far. Because such populations can consist of both proliferating and differentiating cells, the task to assess the overall differentiation status is not trivial and requires a high-throughput, fully automated approach to analyze sufficient data for a statistically significant discrimination to determine cell differentiation. We address the problem of detecting differentiation in a mixed population of proliferating and differentiating cells over time by supervised classification. Using nerve growth factor induced differentiation of PC12 cells, we monitor the changes in cell morphology over 6 days by phase-contrast live-cell imaging. For general applicability, the classification procedure starts out with many features to identify those that maximize discrimination of differentiated and undifferentiated cells and to eliminate features sensitive to systematic measurement artifacts. The resulting image analysis determines the optimal post treatment day for training and achieves a near perfect classification of differentiation, which we confirmed in technically and biologically independent as well as differently designed experiments. Our approach allows to monitor neuronal cell populations repeatedly over days without any interference. It requires only an initial calibration and training step and is thereafter capable to discriminate further experiments. In conclusion, this enables long-term, large-scale studies of cell populations with minimized costs and efforts for detecting effects of external manipulation of neuronal cell differentiation.

Representative members of the bioimage informatics community review the computational steps and some of the primary software tools available to biologists who are acquiring and analyzing microscopy-based digital image data, with a focus on open-source options. Few technologies are more widespread in modern biological laboratories than imaging. Recent advances in optical technologies and instrumentation are providing hitherto unimagined capabilities. Almost all these advances have required the development of software to enable the acquisition, management, analysis and visualization of the imaging data. We review each computational step that biologists encounter when dealing with digital images, the inherent challenges and the overall status of available software for bioimage informatics, focusing on open-source options.

This chapter describes approaches for learning models of subcellular organization from images. The primary utility of these models is expected to be from incorporation into complex simulations of cell behaviors. Most current cell simulations do not consider spatial organization of proteins at all, or treat each organelle type as a single, idealized compartment. The ability to build generative models for all proteins in a proteome and use them for spatially accurate simulations is expected to improve the accuracy of models of cell behaviors. A second use, of potentially equal importance, is expected to be in testing and comparing software for analyzing cell images. The complexity and sophistication of algorithms used in cell-image-based screens and assays (variously referred to as high-content screening, high-content analysis, or high-throughput microscopy) is continuously increasing, and generative models can be used to produce images for testing these algorithms in which the expected answer is known.

Purpose To assess the role of the recombinant bacterial enzyme, glucarpidase (carboxypeptidase-G2), leucovorin, and thymidine in the management and outcome of patients with high-dose methotrexate (HDMTX) –induced nephrotoxicity. Methods Patients with HDMTX-induced nephrotoxicity received one to three doses of intravenous (IV) glucarpidase and leucovorin rescue. The initial cohort (n = 35) also received thymidine by continuous IV infusion. Subsequently, thymidine was restricted to patients with prolonged exposure (> 96 hours) to methotrexate (MTX) or with substantial MTX toxicity at study entry. Plasma MTX, leucovorin, and 5-methyltetrahydrofolate (5-mTHF) concentrations were measured pre- and postglucarpidase. Toxicities were monitored, and logistic regression analysis was used to assess the relationship of baseline characteristics to the development of severe toxicity and death. Results Glucarpidase was administered at a median of 96 hours (receiving thymidine, n = 44) and 66 hours (not receiving thymidine, n = 56) after the start of the MTX infusion. Plasma MTX concentrations decreased within 15 minutes of glucarpidase by 98.7%. Plasma 5-mTHF concentrations also decreased more than 98% after administration of glucarpidase. Of 12 deaths, six were directly attributed to irreversible MTX toxicity. Presence of grade 4 toxicity before administration of glucarpidase, inadequate initial increase in leucovorin dosing, and administration of glucarpidase more than 96 hours after the start of the MTX infusion were associated with development of grade 4 and 5 toxicity. Conclusion Early intervention with the combination of leucovorin and glucarpidase is highly effective in patients who develop HDMTX-induced renal dysfunction. Severe toxicity and mortality occurred in patients in whom glucarpidase rescue was delayed and occurred despite thymidine administration.

The SLIF project combines text-mining and image processing to extract structured information from biomedical literature. SLIF extracts images and their captions from published papers. The captions are automatically parsed for relevant biological entities (protein and cell type names), while the images are classified according to their type (e.g., micrograph or gel). Fluorescence microscopy images are further processed and classified according to the depicted subcellular localization. The results of this process can be queried online using either a user-friendly web-interface or an XML-based web-service. As an alternative to the targeted query paradigm, SLIF also supports browsing the collection based on latent topic models which are derived from both the annotated text and the image data. The SLIF web application, as well as labeled datasets used for training system components, is publicly available at http://slif.cbi.cmu.edu.

The detection of protein–protein interactions through two-hybrid assays has revolutionized our understanding of biology. The remarkable impact of two-hybrid assay platforms derives from their speed, simplicity, and broad applicability. Yet for many organisms, the need to express test proteins in Saccharomyces cerevisiae or Escherichia coli presents a substantial barrier because variations in codon specificity or bias may result in aberrant protein expression. In particular, nonstandard genetic codes are characteristic of several eukaryotic pathogens, for which there are currently no genetically based systems for detection of protein–protein interactions. We have developed a protein–protein interaction assay that is carried out in native host cells by using GFP as the only foreign protein moiety, thus circumventing these problems. We show that interaction can be detected between two protein pairs in both the model yeast S. cerevisiae and the fungal pathogen Candida albicans. We use computational analysis of microscopic images to provide a quantitative and automated assessment of confidence.

Nutlins, the newly developed small molecule antagonists of MDM2, activate p53 and induce apoptosis in cancer cells, offering a novel strategy of chemotherapy. Recent studies have further suggested synergistic effects of nutlins with other chemotherapeutic drugs. However, it is unclear whether nutlins increase or decrease the side effects of these drugs in normal non-malignant cells or tissues. Cisplatin is a widely used chemotherapy drug, which has a major side effect of kidney injury. Here we show that Nutlin-3 protected kidney cells against cisplatin-induced apoptosis. The cytoprotective effects of Nutlin-3 were not related to its regulation of p53 or consequent gene expression during cisplatin treatment. Moreover, the protective effects were shown in MDM2-, MDM4-, or p53-deficient cells. On the other hand, Nutlin-3 suppressed mitochondrial events of apoptosis during cisplatin incubation, including Bax activation and cytochrome c release. Nutlin-3 attenuated cisplatin-induced oligomerization of Bax and Bak but not their interactions with Bcl-XL. In isolated mitochondria, Nutlin-3 inhibited cytochrome c release induced by Ca2+, Bim peptide, and recombinant tBid. Importantly, it blocked both Bax and Bak oligomerization under these conditions. Together, the results have uncovered a new pharmacological function of nutlins, i.e. suppression of Bax and Bak, two critical mediators of apoptosis.

Background Detailed knowledge of the subcellular location of each expressed protein is critical to a full understanding of its function. Fluorescence microscopy, in combination with methods for fluorescent tagging, is the most suitable current method for proteome-wide determination of subcellular location. Previous work has shown that neural network classifiers can distinguish all major protein subcellular location patterns in both 2D and 3D fluorescence microscope images. Building on these results, we evaluate here new classifiers and features to improve the recognition of protein subcellular location patterns in both 2D and 3D fluorescence microscope images. Results We report here a thorough comparison of the performance on this problem of eight different state-of-the-art classification methods, including neural networks, support vector machines with linear, polynomial, radial basis, and exponential radial basis kernel functions, and ensemble methods such as AdaBoost, Bagging, and Mixtures-of-Experts. Ten-fold cross validation was used to evaluate each classifier with various parameters on different Subcellular Location Feature sets representing both 2D and 3D fluorescence microscope images, including new feature sets incorporating features derived from Gabor and Daubechies wavelet transforms. After optimal parameters were chosen for each of the eight classifiers, optimal majority-voting ensemble classifiers were formed for each feature set. Comparison of results for each image for all eight classifiers permits estimation of the lower bound classification error rate for each subcellular pattern, which we interpret to reflect the fraction of cells whose patterns are distorted by mitosis, cell death or acquisition errors. Overall, we obtained statistically significant improvements in classification accuracy over the best previously published results, with the overall error rate being reduced by one-third to one-half and with the average accuracy for single 2D images being higher than 90% for the first time. In particular, the classification accuracy for the easily confused endomembrane compartments (endoplasmic reticulum, Golgi, endosomes, lysosomes) was improved by 5–15%. We achieved further improvements when classification was conducted on image sets rather than on individual cell images. Conclusions The availability of accurate, fast, automated classification systems for protein location patterns in conjunction with high throughput fluorescence microscope imaging techniques enables a new subfield of proteomics, location proteomics. The accuracy and sensitivity of this approach represents an important alternative to low-resolution assignments by curation or sequence-based prediction.

Purpose Research on venous thromboembolism events (VTE), such as deep venous thrombosis (DVT) and pulmonary embolism (PE), in pediatric orthopaedic trauma patients is sparse. We describe the incidence in the USA of VTE associated with pediatric lower extremity orthopaedic trauma, and characterize injury patterns and VTE treatment methods. Methods The Pediatric Health Information System (PHIS) was queried from 2004 to 2013 using ICD-9 codes for lower extremity fractures (pelvis, femur, tibia, ankle, foot) and dislocations (hip, knee, ankle, subtalar) and VTE. Records were queried for age, diagnoses, and VTE treatment. Results During the study period 285,611 clinical encounters reported lower extremity trauma. Of those, 167 patients were simultaneously coded with VTE (99 DVT, 50 PE, 18 combined DVT/PE), to give an incidence of VTE associated with pediatric lower extremity trauma of 0.058 %. Patients were from 39 centers, with an average age of 12.9 years (range 0–19). There were 249 fractures and 21 dislocations, with 25 (15 %) patients sustaining more than one lower extremity injury. The most common fracture locations were the femur/femoral neck (95), tibia/ankle (92), and pelvis (44). 72 % (121/167) of patients were treated with anticoagulation medication, of which the most common was low-molecular-weight heparin (111/167, 66 %). Conclusions The incidence of VTE events associated with lower extremity orthopaedic trauma is 0.058 %. Adolescents and polytrauma patients with injuries of the femur/femoral neck, tibia/ankle, and pelvis are more commonly affected. Low-molecular-weight heparin is commonly used to treat VTE in pediatric and adolescent patients.

Na+,K(+)-ATPase has been observed to partially inhibit acidification of early endosomes by increasing membrane potential, whereas chloride channels have been observed to enhance acidification in endosomes and lysosomes. However, little theoretical analysis of the ways in which different pumps and channels may interact has been carried out. We therefore developed quantitative models of endosomal pH regulation based on thermodynamic considerations. We conclude that 1) both size and shape of endosomes will influence steady-state endosomal pH whenever membrane potential due to the pH gradient limits proton pumping, 2) steady-state pH values similar to those observed in early endosomes of living cells can occur in endosomes containing just H(+)-ATPases and Na+,K(+)-ATPases when low endosomal buffering capacities are present, and 3) inclusion of active chloride channels results in predicted pH values well below those observed in vivo. The results support the separation of endocytic compartments into two classes, those (such as early endosomes) whose acidification is limited by attainment of a certain membrane potential, and those (such as lysosomes) whose acidification is limited by the attainment of a certain pH. The theoretical framework and conclusions described are potentially applicable to other membrane-enclosed compartments that are acidified, such as elements of the Golgi apparatus.

Acidification of endocytosed ligands destined for lysosomes is biphasic, with a rapid drop to pH 6, followed by a slow decrease to pH 5. Continuous measurements of transferrin acidification have confirmed that the pH minimum in early (presorting) endosomes is approximately pH 6. On the basis of measurements of endosomal acidification in vitro, it has been proposed that the pH in the early endosome is limited by the internalization of the Na+,K+-ATPase, which generates an interior-positive membrane potential in this compartment [Fuchs, R., Schmid, S. & Mellman, I. (1989) Proc. Natl. Acad. Sci. USA 86, 539-543]. We present two lines of evidence that strongly implicate the Na+,K+-ATPase as a major regulatory element of endocytic pH in vivo. First, ouabain, a specific inhibitor of the Na+,K+-ATPase, interferes with the regulation of acidification in early endocytic compartments. Transferrin is normally rapidly acidified to pH 6.0-6.2, followed by alkalinization during recycling. In the presence of ouabain, the minimum pH of transferrin-containing endosomes decreases from 6.0-6.2 to less than 5.3. Second, ouabain eliminates the resistance to both the growth inhibitory and vacuologenic effects of chloroquine in the lysosomal acidification defective cell line CHL60-64. The phenotype of this cell line is consistent with a defect in the removal or inactivation of the early acidification regulatory elements from the late endocytic compartments. The ouabain data suggest that the defect in this cell line is due to improper localization of the Na+,K+-ATPase. A model for pH regulation and vacuolation by weak bases is discussed.

The properties of chromatin containing newly synthesized DNA and protein were investigated. A fraction of chromatin enriched in newly replicated DNA was isolated by means of its increased density in metrizamide relative to bulk chromatin. The DNA of this fraction appeared to be packaged into nucleosomes but at a reduced nucleosomal spacing. Although pulse-labeled DNA was present in this dense fraction, nucleosomes labeled with short pulses of arginine or acetate were of normal density. The data presented are consistent with the conclusion that newly replicated DNA is associated with preexisting histones in a short-lived, compact structure whereas newly synthesized histones are deposited at normal spacing some distance from the replication fork.

Oligodeoxyribonucleotides complementary to the DNA of the wild type (wt) bacteriophage phi chi 174 have been synthesized by the phosphotriester method. The oligomers, 11, 14, and 17 bases long, are complementary to the region of the DNA which accounts for the am-3 point mutation. When hybridized to am-3 DNA, the oligonucleotides form duplexes with a single base pair mismatch. The thermal stability of the duplexes formed between wt and am-3 DNAs has been measured. The am-3 DNA:oligomer duplexes dissociate at a temperature about 10 degrees C lower than the corresponding wt DNA:oligomer duplexes. This dramatic decrease in thermal stability due to a single mismatch makes it possible to eliminate the formation of the mismatched duplexes by the appropriate choice of hybridization temperature. These results are discussed with respect to the use of oligonucleotides as probes for the isolation of specific cloned DNA sequences.

Background: Drug discovery and development has been aided by high throughput screening methods that detect compound effects on a single target. However, when using focused initial screening, undesirable secondary effects are often detected late in the development process after significant investment has been made. An alternative approach would be to screen against undesired effects early in the process, but the number of possible secondary targets makes this prohibitively expensive. Results: This paper describes methods for making this global approach practical by constructing predictive models for many target responses to many compounds and using them to guide experimentation. We demonstrate for the first time that by jointly modeling targets and compounds using descriptive features and using active machine learning methods, accurate models can be built by doing only a small fraction of possible experiments. The methods were evaluated by computational experiments using a dataset of 177 assays and 20,000 compounds constructed from the PubChem database. Conclusions: An average of nearly 60% of all hits in the dataset were found after exploring only 3% of the experimental space which suggests that active learning can be used to enable more complete characterization of compound effects than otherwise affordable. The methods described are also likely to find widespread application outside drug discovery, such as for characterizing the effects of a large number of compounds or inhibitory RNAs on a large number of cell or tissue phenotypes.