Abs are central to malaria immunity, which is only acquired after years of exposure to Plasmodium falciparum (Pf). Despite the enormous worldwide burden of malaria, the targets of protective Abs and the basis of their inefficient acquisition are unknown. Addressing these knowledge gaps could accelerate malaria vaccine development. To this end, we developed a protein microarray containing ∼23% of the Pf 5,400-protein proteome and used this array to probe plasma from 220 individuals between the ages of 2-10 years and 18-25 years in Mali before and after the 6-month malaria season. Episodes of malaria were detected by passive surveillance over the 8-month study period. Ab reactivity to Pf proteins rose dramatically in children during the malaria season; however, most of this response appeared to be short-lived based on cross-sectional analysis before the malaria season, which revealed only modest incremental increases in Ab reactivity with age. Ab reactivities to 49 Pf proteins measured before the malaria season were significantly higher in 8-10-year-old children who were infected with Pf during the malaria season but did not experience malaria (n = 12) vs. those who experienced malaria (n = 29). This analysis also provided insight into patterns of Ab reactivity against Pf proteins based on the life cycle stage at which proteins are expressed, subcellular location, and other proteomic features. This approach, if validated in larger studies and in other epidemiological settings, could prove to be a useful strategy for better understanding fundamental properties of the human immune response to Pf and for identifying previously undescribed vaccine targets.antigen discovery | naturally acquired immunity | Plasmodium falciparum malaria | prospective cohort study | Mali

pfbaldi@ics.uci.edu

Understanding the way in which the immune system responds to infection is central to the development of vaccines and many diagnostics. To provide insight into this area, we fabricated a protein microarray containing 1,205 Burkholderia pseudomallei proteins, probed it with 88 melioidosis patient sera, and identified 170 reactive antigens. This subset of antigens was printed on a smaller array and probed with a collection of 747 individual sera derived from 10 patient groups including melioidosis patients from Northeast Thailand and Singapore, patients with different infections, healthy individuals from the USA, and from endemic and nonendemic regions of Thailand. We identified 49 antigens that are significantly more reactive in melioidosis patients than healthy people and patients with other types of bacterial infections. We also identified 59 cross-reactive antigens that are equally reactive among all groups, including healthy controls from the USA. Using these results we were able to devise a test that can classify melioidosis positive and negative individuals with sensitivity and specificity of 95% and 83%, respectively, a significant improvement over currently available diagnostic assays. Half of the reactive antigens contained a predicted signal peptide sequence and were classified as outer membrane, surface structures or secreted molecules, and an additional 20% were associated with pathogenicity, adaptation or chaperones. These results show that microarrays allow a more comprehensive analysis of the immune response on an antigen-specific, patient-specific, and population-specific basis, can identify serodiagnostic antigens, and contribute to a more detailed understanding of immunogenicity to this pathogen.antigen discovery ͉ melioidosis ͉ diagnostic ͉ antigen prediction

Humans and other animals with Lyme borreliosis produce antibodies to a number of components of the agent Borrelia burgdorferi, but a full accounting of the immunogens during natural infections has not been achieved. Employing a protein array produced in vitro from 1,292 DNA fragments representing ϳ80% of the genome, we compared the antibody reactivities of sera from patients with early or later Lyme borreliosis to the antibody reactivities of sera from controls. Overall, ϳ15% of the open reading frame (ORF) products ( Infectious disease research has advanced rapidly with the accumulation of whole-genome sequences of pathogens and the subsequent use of genome-wide DNA microarrays to study gene expression. Equipped with arrays in different formats, investigators have identified different genes in a variety of pathogens that are more highly expressed in host animals or under in vitro conditions mimicking the in vivo environment. With few exceptions (27), these array studies have been performed with experimental animals, usually rodents, and in laboratory settings. Less is known about the proteins that are expressed during natural infections of humans or other host animals. Detection of a specific antibody during the course of infection is indirect evidence of in vivo expression by the pathogen. But the use of this approach to study large numbers of proteins has been largely limited to one-dimensional and two-dimensional gel electrophoresis of whole cells having an in vitro origin, followed by identification of the more abundant antigens by partial amino acid sequencing of reactive bands or spots and then searches of the databases (22,23,38,44,52,60,66).An alternative to using the pathogen itself as the source of the proteins is to produce recombinant polypeptides based on the deduced open reading frames (ORFs) of the pathogen's genome and to determine whether these polypeptides are antibody targets (11,61). A potential shortcoming of using this approach with cells, such as Escherichia coli or yeast (Saccharomyces cerevisiae) cells, is that some foreign proteins may not be expressed or the quantity may be insufficient. Felgner and coinvestigators increased the success rate for expression and decreased the cost with a high-throughput, cell-free, coupled transcription-translation system (26,29,30). Hundreds to thousands of individual recombinant proteins are printed on chips, which are then used to capture antibodies present in serum from infected individuals and other animals. The amount of captured antibody is quantified using a labeled secondary antibody. Whole-proteome microarrays have been employed in studies of experimental Francisella tularensis infections in laboratory mice (35,57) and of immune responses of humans to immunization with live vaccinia virus (26,29,31). McKevitt et al. (61) and Brinkmann et al. (11) used the enzyme-linked immunosorbent assay (ELISA) format to study the binding of antibodies of experimentally infected rabbits and people with syphilis to a nearly complete representation of the OR...

Being able to predict the course of arbitrary chemical reactions is essential to the theory and applications of organic chemistry. Approaches to the reaction prediction problems can be organized around three poles corresponding to: (1) physical laws; (2) rule-based expert systems; and (3) inductive machine learning. Previous approaches at these poles respectively are not high-throughput, are not generalizable or scalable, or lack sufficient data and structure to be implemented. We propose a new approach to reaction prediction utilizing elements from each pole. Using a physically inspired conceptualization, we describe single mechanistic reactions as interactions between coarse approximations of molecular orbitals (MOs) and use topological and physicochemical attributes as descriptors. Using an existing rule-based system (Reaction Explorer), we derive a restricted chemistry dataset consisting of 1630 full multi-step reactions with 2358 distinct starting materials and intermediates, associated with 2989 productive mechanistic steps and 6.14 million unproductive mechanistic steps. And from machine learning, we pose identifying productive mechanistic steps as a statistical ranking, information retrieval, problem: given a set of reactants and a description of conditions, learn a ranking model over potential filled-to-unfilled MO interactions such that the top ranked mechanistic steps yield the major products. The machine learning implementation follows a two-stage approach, in which we first train atom level reactivity filters to prune 94.00% of non-productive reactions with a 0.01% error rate. Then, we train an ensemble of ranking models on pairs of interacting MOs to learn a relative productivity function over mechanistic steps in a given system. Without the use of explicit transformation patterns, the ensemble perfectly ranks the productive mechanism at the top 89.05% of the time, rising to 99.86% of the time when the top four are considered. Furthermore, the system is generalizable, making reasonable predictions over reactants and conditions which the rule-based expert does not handle. A web interface to the machine learning based mechanistic reaction predictor is accessible through our chemoinformatics portal (http://cdb.ics.uci.edu) under the Toolkits section.

The development of an effective malaria vaccine remains a global public health priority. Less than 0.5% of the Plasmodium falciparum genome has been assessed as potential vaccine targets and candidate vaccines have been based almost exclusively on single antigens. It is possible that the failure to develop a malaria vaccine despite decades of effort might be attributed to this historic focus. To advance malaria vaccine development, we have fabricated protein microarrays representing 23% of the entire P. falciparum proteome and have probed these arrays with plasma from subjects with sterile protection or no protection after experimental immunization with radiation attenuated P. falciparum sporozoites. A panel of 19 pre-erythrocytic stage antigens was identified as strongly associated with sporozoite-induced protective immunity; 16 of these antigens were novel and 85% have been independently identified in sporozoite and/or liver stage proteomic or transcriptomic data sets. Reactivity to any individual antigen did not correlate with protection but there was a highly significant difference in the cumulative signal intensity between protected and not protected individuals. Functional annotation indicates that most of these signature proteins are involved in cell cycle/DNA processing and protein synthesis. In addition, 21 novel blood-stage specific antigens were identified. Our data provide the first evidence that sterile protective immunity against malaria is directed against a panel of novel P. falciparum antigens rather than one antigen in isolation. These results have important implications for vaccine development, suggesting that an efficacious malaria vaccine should be multivalent and targeted at a select panel of key antigens, many of which have not been previously

Proposing reasonable mechanisms and predicting the course of chemical reactions is important to the practice of organic chemistry. Approaches to reaction prediction have historically used obfuscating representations and manually encoded patterns or rules. Here we present ReactionPredictor, a machine learning approach to reaction prediction that models elementary, mechanistic reactions as interactions between approximate molecular orbitals (MOs). A training data set of productive reactions known to occur at reasonable rates and yields and verified by inclusion in the literature or textbooks is derived from an existing rule-based system and expanded upon with manual curation from graduate level textbooks. Using this training data set of complex polar, hypervalent, radical, and pericyclic reactions, a two-stage machine learning prediction framework is trained and validated. In the first stage, filtering models trained at the level of individual MOs are used to reduce the space of possible reactions to consider. In the second stage, ranking models over the filtered space of possible reactions are used to order the reactions such that the productive reactions are the top ranked. The resulting model, ReactionPredictor, perfectly ranks polar reactions 78.1% of the time and recovers all productive reactions 95.7% of the time when allowing for small numbers of errors. Pericyclic and radical reactions are perfectly ranked 85.8% and 77.0% of the time, respectively, rising to >93% recovery for both reaction types with a small number of allowed errors. Decisions about which of the polar, pericyclic, or radical reaction type ranking models to use can be made with >99% accuracy. Finally, for multistep reaction pathways, we implement the first mechanistic pathway predictor using constrained tree-search to discover a set of reasonable mechanistic steps from given reactants to given products. Webserver implementations of both the single step and pathway versions of ReactionPredictor are available via the chemoinformatics portal http://cdb.ics.uci.edu/.

The Bayesian regularization method for high-throughput differential analysis, described in Baldi and Long (A Bayesian framework for the analysis of microarray expression data: regularized t-test and statistical inferences of gene changes. Bioinformatics 2001: 17: 509-519) and implemented in the Cyber-T web server, is one of the most widely validated. Cyber-T implements a t-test using a Bayesian framework to compute a regularized variance of the measurements associated with each probe under each condition. This regularized estimate is derived by flexibly combining the empirical measurements with a prior, or background, derived from pooling measurements associated with probes in the same neighborhood. This approach flexibly addresses problems associated with low replication levels and technology biases, not only for DNA microarrays, but also for other technologies, such as protein arrays, quantitative mass spectrometry and next-generation sequencing (RNA-seq). Here we present an update to the Cyber-T web server, incorporating several useful new additions and improvements. Several preprocessing data normalization options including logarithmic and (Variance Stabilizing Normalization) VSN transforms are included. To augment two-sample t-tests, a one-way analysis of variance is implemented. Several methods for multiple tests correction, including standard frequentist methods and a probabilistic mixture model treatment, are available. Diagnostic plots allow visual assessment of the results. The web server provides comprehensive documentation and example data sets. The Cyber-T web server, with R source code and data sets, is publicly available at http://cybert.ics.uci.edu/.