A proteome-wide screen of Campylobacter jejuni using protein microarrays identifies novel and conformational antigens.

Campylobacter jejuni (C. jejuni) is a foodborne intestinal pathogen and major cause of gastroenteritis worldwide. C. jejuni proteins that are immunogenic have been sought for their potential use in the development of biomarkers, diagnostic assays, or subunit vaccines for humans or livestock. To identify new immunogenic C. jejuni proteins, we used a native protein microarray approach. A protein chip, with over 1400 individually purified GST-tagged C. jejuni proteins, representing over 86% of the proteome, was constructed to screen for antibody titers present in test sera raised against whole C. jejuni cells. Dual detection of GST signals was incorporated as a way of normalizing the variation of protein concentrations contributing to the antibody staining intensities. We detected strong signals to 102 C. jejuni antigens. In addition to antigens recognized by antiserum raised against C. jejuni, parallel experiments were conducted to identify antigens cross-reactive to antiserum raised against various serotypes of E. coli or Salmonella or to healthy human sera. This led to the identification of 34 antigens specifically recognized by the C. jejuni antiserum, only four of which were previously known. The chip approach also allowed identification of conformational antigens. We demonstrate in the case of Cj1621 that antigen signals are lost to denaturing conditions commonly used in other approaches to identify immunogens. Antigens identified in this study include those possessing sequence features indicative of cell surface localization, as well as those that do not. Together, our results indicate that the unbiased chip-based screen can help reveal the full repertoire of host antibodies against microbial proteomes.

A novel ER-microtubule-binding protein, ERLIN2, stabilizes Cyclin B1 and regulates cell cycle progression.

The gene encoding endoplasmic reticulum (ER) lipid raft-associated protein 2 (ERLIN2) is amplified in human breast cancers. ERLIN2 gene mutations were also found to be associated with human childhood progressive motor neuron diseases. Yet, an understanding of the physiological function and mechanism for ERLIN2 remains elusive. In this study, we reveal that ERLIN2 is a spatially and temporally regulated ER-microtubule-binding protein that has an important role in cell cycle progression by interacting with and stabilizing the mitosis-promoting factors. Whereas ERLIN2 is highly expressed in aggressive human breast cancers, during normal development ERLIN2 is expressed at the postnatal stage and becomes undetectable in adulthood. ERLIN2 interacts with the microtubule component α-tubulin, and this interaction is maximal during the cell cycle G2/M phase where ERLIN2 simultaneously interacts with the mitosis-promoting complex Cyclin B1/Cdk1. ERLIN2 facilitates K63-linked ubiquitination and stabilization of Cyclin B1 protein in G2/M phase. Downregulation of ERLIN2 results in cell cycle arrest, represses breast cancer proliferation and malignancy and increases sensitivity of breast cancer cells to anticancer drugs. In summary, our study revealed a novel ER-microtubule-binding protein, ERLIN2, which interacts with and stabilizes mitosis-promoting factors to regulate cell cycle progression associated with human breast cancer malignancy.

A role for Drosophila Cyclin J in oogenesis revealed by genetic interactions with the piRNA pathway.

Cyclin J (CycJ) is a poorly characterized member of the Cyclin superfamily of cyclin-dependent kinase regulators, many of which regulate the cell cycle or transcription. Although CycJ is conserved in metazoans its cellular function has not been identified and no mutant defects have been described. In Drosophila, CycJ transcript is present primarily in ovaries and very early embryos, suggesting a role in one or both of these tissues. The CycJ gene (CycJ) lies immediately downstream of armitage (armi), a gene involved in the Piwi-associated RNA (piRNA) pathways that are required for silencing transposons in the germline and adjacent somatic cells. Mutations in armi result in oogenesis defects but a role for CycJ in oogenesis has not been defined. Here we assessed oogenesis in CycJ mutants in the presence or absence of mutations in armi or other piRNA pathway genes. CycJ null ovaries appeared normal, indicating that CycJ is not essential for oogenesis under normal conditions. In contrast, armi null ovaries produced only two egg chambers per ovariole and the eggs had severe axis specification defects, as observed previously for armi and other piRNA pathway mutants. Surprisingly, the CycJ armi double mutant failed to produce any mature eggs. The double null ovaries generally had only one egg chamber per ovariole and the egg chambers frequently contained an overabundance of differentiated germline cells. Production of these compound egg chambers could be suppressed with CycJ transgenes but not with mutations in the checkpoint gene mnk, which suppress oogenesis defects in armi mutants. The CycJ null showed similar genetic interactions with the germline and somatic piRNA pathway gene piwi, and to a lesser extent with aubergine (aub), a member of the germline-specific piRNA pathway. The strong genetic interactions between CycJ and piRNA pathway genes reveal a role for CycJ in early oogenesis. Our results suggest that CycJ is required to regulate egg chamber production or maturation when piRNA pathways are compromised.

Integrating the interactome and the transcriptome of Drosophila.

BACKGROUND: Networks of interacting genes and gene products mediate most cellular and developmental processes. High throughput screening methods combined with literature curation are identifying many of the protein-protein interactions (PPI) and protein-DNA interactions (PDI) that constitute these networks. Most of the detection methods, however, fail to identify the in vivo spatial or temporal context of the interactions. Thus, the interaction data are a composite of the individual networks that may operate in specific tissues or developmental stages. Genome-wide expression data may be useful for filtering interaction data to identify the subnetworks that operate in specific spatial or temporal contexts. Here we take advantage of the extensive interaction and expression data available for Drosophila to analyze how interaction networks may be unique to specific tissues and developmental stages. RESULTS: We ranked genes on a scale from ubiquitously expressed to tissue or stage specific and examined their interaction patterns. Interestingly, ubiquitously expressed genes have many more interactions among themselves than do non-ubiquitously expressed genes both in PPI and PDI networks. While the PDI network is enriched for interactions between tissue-specific transcription factors and their tissue-specific targets, a preponderance of the PDI interactions are between ubiquitous and non-ubiquitously expressed genes and proteins. In contrast to PDI, PPI networks are depleted for interactions among tissue- or stage- specific proteins, which instead interact primarily with widely expressed proteins. In light of these findings, we present an approach to filter interaction data based on gene expression levels normalized across tissues or developmental stages. We show that this filter (the percent maximum or pmax filter) can be used to identify subnetworks that function within individual tissues or developmental stages. CONCLUSIONS: These observations suggest that protein networks are frequently organized into hubs of widely expressed proteins to which are attached various tissue- or stage-specific proteins. This is consistent with earlier analyses of human PPI data and suggests a similar organization of interaction networks across species. This organization implies that tissue or stage specific networks can be best identified from interactome data by using filters designed to include both ubiquitously expressed and specifically expressed genes and proteins.

Identification of new protein interactions between dengue fever virus and its hosts, human and mosquito.

The four divergent serotypes of dengue virus are the causative agents of dengue fever, dengue hemorrhagic fever and dengue shock syndrome. About two-fifths of the world's population live in areas where dengue is prevalent, and thousands of deaths are caused by the viruses every year. Dengue virus is transmitted from one person to another primarily by the yellow fever mosquito, Aedes aegypti. Recent studies have begun to define how the dengue viral proteins interact with host proteins to mediate viral replication and pathogenesis. A combined analysis of these studies, however, suggests that many virus-host protein interactions remain to be identified, especially for the mosquito host. In this study, we used high-throughput yeast two-hybrid screening to identify mosquito and human proteins that physically interact with dengue proteins. We tested each identified host protein against the proteins from all four serotypes of dengue to identify interactions that are conserved across serotypes. We further confirmed many of the interactions using co-affinity purification assays. As in other large-scale screens, we identified some previously detected interactions and many new ones, moving us closer to a complete host - dengue protein interactome. To help summarize and prioritize the data for further study, we combined our interactions with other published data and identified a subset of the host-dengue interactions that are now supported by multiple forms of evidence. These data should be useful for understanding the interplay between dengue and its hosts and may provide candidates for drug targets and vector control strategies.

Top-k similar graph matching using TraM in biological networks.

Many emerging database applications entail sophisticated graph-based query manipulation, predominantly evident in large-scale scientific applications. To access the information embedded in graphs, efficient graph matching tools and algorithms have become of prime importance. Although the prohibitively expensive time complexity associated with exact subgraph isomorphism techniques has limited its efficacy in the application domain, approximate yet efficient graph matching techniques have received much attention due to their pragmatic applicability. Since public domain databases are noisy and incomplete in nature, inexact graph matching techniques have proven to be more promising in terms of inferring knowledge from numerous structural data repositories. In this paper, we propose a novel technique called TraM for approximate graph matching that off-loads a significant amount of its processing on to the database making the approach viable for large graphs. Moreover, the vector space embedding of the graphs and efficient filtration of the search space enables computation of approximate graph similarity at a throw-away cost. We annotate nodes of the query graphs by means of their global topological properties and compare them with neighborhood biased segments of the datagraph for proper matches. We have conducted experiments on several real data sets, and have demonstrated the effectiveness and efficiency of the proposed method

High-throughput yeast two-hybrid screening.

Charting the interactions among proteins is essential for understanding biological processes. While a number of complementary technologies for detecting protein interactions are available, the yeast two-hybrid system is one of the few that have been successfully scaled up. Two-hybrid screens have been used to construct extensive protein interaction maps for humans and several model organisms, and these maps have proven invaluable for studies on a variety of biological systems. These maps, however, have not come close to covering all proteins or interactions detectable by yeast two-hybrid. This is due in part to the difficulty of using library screening methods to sample all possible binary combinations of proteins. Ideally, every binary pair of proteins would be tested individually to ensure that every detectable interaction is identified. For organisms with large proteomes, however, this is not economically feasible and instead efficient pooling schemes must be implemented. The high-throughput two-hybrid screening methods presented here are designed to efficiently maximize coverage for selected sets of proteins or entire proteomes. We present two high-throughput screening protocols. Both methods are designed to identify interactors for any number of bait proteins expressed as DNA-binding domain (BD) fusions. The choice of which protocol to use depends largely on the nature of the available library of proteins fused to an activation domain (AD). The first protocol is appropriate for screening a library of AD clones, such as a cDNA library, a domain library, or a large pool of AD clones. By contrast, the second protocol is appropriate for screening a large array of individual sequence-verified AD clones. This protocol screens small pools of AD clones from the array in a two-phase scheme. Although the methods presented were developed using the LexA version of the yeast two-hybrid system, we include notes as appropriate to accommodate users of other versions.

Assigning confidence scores to protein-protein interactions.

Screens for protein-protein interactions using assays like the yeast two-hybrid system have generated volumes of useful data. The protein interactions from these screens have been used to develop a better understanding of the functions of individual proteins, regulatory pathways, molecular machines, and entire biological systems. The value of this data, however, is limited by the inherent frequency of false positives that arise in most protein interaction screens. Appreciable numbers of false positives can crop up in both low-throughput and high-throughput screens, and even in screens that employ stringent criteria for defining a positive. A number of classification systems have been used to help distinguish false positives from biologically relevant true positives. This chapter describes a system for assigning a confidence score to each interaction based on the probability that it is a true positive. Such confidence scores can be used to prioritize interactions for validation. The scores are also useful for network analysis methods that take advantage of probabilistic edge weights. The scoring method does not rely on gold standard datasets of reliable true positives and true negatives, and thus circumvents the challenges associated with obtaining such datasets. Moreover, the scoring method uses data features that are largely assay-independent, making it useful for interactions obtained from a variety of different technologies and screening methods.

Interaction trap/two-hybrid system to identify interacting proteins.

The yeast two-hybrid method (or interaction trap) is a powerful technique for detecting protein interactions. The procedure is performed using transcriptional activation of a dual reporter system in yeast to identify interactions between a protein of interest (the bait protein) and the candidate proteins for interaction. The method can be used to screen a protein library for interactions with a bait protein or to test for association between proteins that are expected to interact based on prior evidence. Interaction mating facilitates the screening of a library with multiple bait proteins.

Proteomic and functional genomic landscape of receptor tyrosine kinase and ras to extracellular signal-regulated kinase signaling.

Characterizing the extent and logic of signaling networks is essential to understanding specificity in such physiological and pathophysiological contexts as cell fate decisions and mechanisms of oncogenesis and resistance to chemotherapy. Cell-based RNA interference (RNAi) screens enable the inference of large numbers of genes that regulate signaling pathways, but these screens cannot provide network structure directly. We describe an integrated network around the canonical receptor tyrosine kinase (RTK)-Ras-extracellular signal-regulated kinase (ERK) signaling pathway, generated by combining parallel genome-wide RNAi screens with protein-protein interaction (PPI) mapping by tandem affinity purification-mass spectrometry. We found that only a small fraction of the total number of PPI or RNAi screen hits was isolated under all conditions tested and that most of these represented the known canonical pathway components, suggesting that much of the core canonical ERK pathway is known. Because most of the newly identified regulators are likely cell type- and RTK-specific, our analysis provides a resource for understanding how output through this clinically relevant pathway is regulated in different contexts. We report in vivo roles for several of the previously unknown regulators, including CG10289 and PpV, the Drosophila orthologs of two components of the serine/threonine-protein phosphatase 6 complex; the Drosophila ortholog of TepIV, a glycophosphatidylinositol-linked protein mutated in human cancers; CG6453, a noncatalytic subunit of glucosidase II; and Rtf1, a histone methyltransferase.

Antagonists of anaphase-promoting complex (APC)-2-cell cycle and apoptosis regulatory protein (CARP)-1 interaction are novel regulators of cell growth and apoptosis.

CARP-1/CCAR1, a perinuclear phosphoprotein, is a regulator of cell growth and apoptosis signaling. Although CARP-1 is a regulator of chemotherapy-dependent apoptosis, it is also a part of the NF-κB proteome and a co-activator of steroid/thyroid nuclear receptors as well as ß-catenin signaling. Our yeast two-hybrid screen revealed CARP-1 binding with the anaphase-promoting complex/cyclosome E3 ubiquitin ligase component APC-2 protein. CARP-1 also binds with anaphase-promoting complex/cyclosome co-activators Cdc20 and Cdh1. Following mapping of the minimal epitopes involved in CARP-1 binding with APC-2, a fluorescence polarization assay was established that indicated a dissociation constant (K(d)) of 480 nm for CARP-1/APC-2 binding. Fluorescence polarization assay-based high throughput screening of a chemical library yielded several small molecule antagonists of CARP-1/APC-2 binding, termed CARP-1 functional mimetics. CFM-4 (1(2-chlorobenzyl)-5'-phenyl-3'H-spiro[indoline-3,2'-[1,3,4]thiadiazol]-2-one), a lead compound, binds with and stimulates CARP-1 expression. CFM-4 prevents CARP-1 binding with APC-2, causes G(2)M cell cycle arrest, and induces apoptosis with an IC(50) range of 10-15 µm. Apoptosis signaling by CFM-4 involves activation of caspase-8 and -9 and caspase-mediated ubiquitin-proteasome pathway-independent loss of cyclin B1 and Cdc20 proteins. Depletion of CARP-1, however, interferes with CFM-4-dependent cell growth inhibition, activation of caspases, and apoptosis. Because CFM-4 also suppresses growth of drug-resistant human breast cancer cells without affecting the growth of human breast epithelial MCF-10A cells, elevating CARP-1 by CFM-4 and consequent apoptosis could in principle be exploited to further elucidate, and perhaps effectively target, often deregulated cell cycle pathways in pathological conditions, including cancer.

A protein network-guided screen for cell cycle regulators in Drosophila.

BACKGROUND: Large-scale RNAi-based screens are playing a critical role in defining sets of genes that regulate specific cellular processes. Numerous screens have been completed and in some cases more than one screen has examined the same cellular process, enabling a direct comparison of the genes identified in separate screens. Surprisingly, the overlap observed between the results of similar screens is low, suggesting that RNAi screens have relatively high levels of false positives, false negatives, or both. RESULTS: We re-examined genes that were identified in two previous RNAi-based cell cycle screens to identify potential false positives and false negatives. We were able to confirm many of the originally observed phenotypes and to reveal many likely false positives. To identify potential false negatives from the previous screens, we used protein interaction networks to select genes for re-screening. We demonstrate cell cycle phenotypes for a significant number of these genes and show that the protein interaction network is an efficient predictor of new cell cycle regulators. Combining our results with the results of the previous screens identified a group of validated, high-confidence cell cycle/cell survival regulators. Examination of the subset of genes from this group that regulate the G1/S cell cycle transition revealed the presence of multiple members of three structurally related protein complexes: the eukaryotic translation initiation factor 3 (eIF3) complex, the COP9 signalosome, and the proteasome lid. Using a combinatorial RNAi approach, we show that while all three of these complexes are required for Cdk2/Cyclin E activity, the eIF3 complex is specifically required for some other step that limits the G1/S cell cycle transition. CONCLUSIONS: Our results show that false positives and false negatives each play a significant role in the lack of overlap that is observed between similar large-scale RNAi-based screens. Our results also show that protein network data can be used to minimize false negatives and false positives and to more efficiently identify comprehensive sets of regulators for a process. Finally, our data provides a high confidence set of genes that are likely to play key roles in regulating the cell cycle or cell survival.

Interaction trap/two-hybrid system to identify interacting proteins.

The yeast two-hybrid method (or interaction trap) is a powerful technique for detecting protein interactions. The procedure is performed using transcriptional activation of a dual reporter system in yeast to identify interactions between a protein of interest (the bait protein) and the candidate proteins for interaction. The method can be used to screen a protein library for interactions with a bait protein or to test for association between proteins that are expected to interact based on prior evidence. Interaction mating facilitates the screening of a library with multiple bait proteins.

DroID 2011: a comprehensive, integrated resource for protein, transcription factor, RNA and gene interactions for Drosophila.

DroID (http://droidb.org/), the Drosophila Interactions Database, is a comprehensive public resource for Drosophila gene and protein interactions. DroID contains genetic interactions and experimentally detected protein-protein interactions curated from the literature and from external databases, and predicted protein interactions based on experiments in other species. Protein interactions are annotated with experimental details and periodically updated confidence scores. Data in DroID is accessible through user-friendly, intuitive interfaces that allow simple or advanced searches and graphical visualization of interaction networks. DroID has been expanded to include interaction types that enable more complete analyses of the genetic networks that underlie biological processes. In addition to protein-protein and genetic interactions, the database now includes transcription factor-gene and regulatory RNA-gene interactions. In addition, DroID now has more gene expression data that can be used to search and filter interaction networks. Orthologous gene mappings of Drosophila genes to other organisms are also available to facilitate finding interactions based on gene names and identifiers for a number of common model organisms and humans. Improvements have been made to the web and graphical interfaces to help biologists gain a comprehensive view of the interaction networks relevant to the genes and systems that they study.

Why cyclin Y? A highly conserved cyclin with essential functions.

Cyclin Y is one of the most highly conserved members of the cyclin superfamily of proteins, which are famous for their crucial roles in regulating the cell cycle and transcription. Despite this high degree of conservation, very little was known about Cyclin Y function prior to a handful of studies published in this past year. Cyclins typically function by activating cyclin-dependent kinases (Cdks) and one insight has come from the identification of a Cdk that is activated by Cyclin Y. Yeast two-hybrid data first linked Cyclin Y with Cdk14, known as Eip63E in Drosophila or PFTAIRE1 in vertebrates. In Drosophila, both Cyclin Y and Eip63E are essential at many stages of development, from embryogenesis to metamorphosis and null mutants show a similar spectrum of developmental defects. In cultured cells, Cyclin Y and Eip63E were shown to phosphorylate the Wg/Wnt co-receptor Arrow/LRP6 in a ligand-independent manner. Eip63E is recruited to LRP6 at the plasma membrane by interacting with Cyclin Y, which is tethered to the membrane through an N-terminal myristoylation. Cyclin Y-dependent LRP6 phosphorylation appears to prime the receptor for subsequent ligand-dependent phosphorylation and activation of the canonical Wnt signaling pathway. Interestingly, Wnt receptor phosphorylation and signaling is maximal in G2/M when Cyclin Y is at its highest levels, suggesting that Cyclin Y may serve to entrain Wnt signaling to the cell cycle. Given the wide range of roles for Wnt signaling during development, these studies may help explain why Cyclin Y is required at several developmental stages and in turn why these proteins are so well conserved in metazoans.

Genetic diversity in Campylobacter jejuni is associated with differential colonization of broiler chickens and C57BL/6J IL10-deficient mice.

Previous studies have demonstrated that Campylobacter jejuni, the leading causative agent of bacterial food-borne disease in the USA, exhibits high-frequency genetic variation that is associated with changes in cell-surface antigens and ability to colonize chickens. To expand our understanding of the role of genetic diversity in the disease process, we analysed the ability of three C. jejuni human disease isolates (strains 11168, 33292 and 81-176) and genetically marked derivatives to colonize Ross 308 broilers and C57BL/6J IL10-deficient mice. C. jejuni colonized broilers at much higher efficiency (all three strains, 23 of 24 broilers) than mice (11168 only, 8 of 24 mice). C. jejuni 11168 genetically marked strains colonized mice at very low efficiency (2 of 42 mice); however, C. jejuni reisolated from mice colonized both mice and broilers at high efficiency, suggesting that this pathogen can adapt genetically in the mouse. We compared the genome composition in the three wild-type C. jejuni strains and derivatives by microarray DNA/DNA hybridization analysis; the data demonstrated a high degree of genetic diversity in three gene clusters associated with synthesis and modification of the cell-surface structures capsule, flagella and lipo-oligosaccharide. Finally, we analysed the frequency of mutation in homopolymeric tracts associated with the contingency genes wlaN (GC tract) and flgR (AT tracts) in culture and after passage through broilers and mice. C. jejuni adapted genetically in culture at high frequency and the degree of genetic diversity was increased by passage through broilers but was nearly eliminated in the gastrointestinal tract of mice. The data suggest that the broiler gastrointestinal tract provides an environment which promotes outgrowth and genetic variation in C. jejuni; the enhancement of genetic diversity at this location may contribute to its importance as a human disease reservoir.

Cyclin Y is a novel conserved cyclin essential for development in Drosophila.

The Drosophila gene CG14939 encodes a member of a highly conserved family of cyclins, the Y-type cyclins, which have not been functionally characterized in any organism. Here we report the generation and phenotypic characterization of a null mutant of CG14939, which we rename Cyclin Y (CycY). We show that the null mutant, CycY(E8), is homozygous lethal with most mutant animals arresting during pupal development. The mutant exhibits delayed larval growth and major developmental defects during metamorphosis, including impaired gas bubble translocation, head eversion, leg elongation, and adult tissue growth. Heat-shock-induced expression of CycY at different times during development resulted in variable levels of rescue, the timing of which suggests a key function for zygotic CycY during the transition from third instar larvae to prepupae. CycY also plays an essential role during embryogenesis since zygotic null embryos from null mothers fail to hatch into first instar larvae. We provide evidence that the CycY protein (CycY) interacts with Eip63E, a cyclin-dependent kinase (Cdk) for which no cyclin partner had previously been identified. Like CycY, the Eip63E gene has essential functions during embryogenesis, larval development, and metamorphosis. Our data suggest that CycY/Eip63E form a cyclin/Cdk complex that is essential for several developmental processes.

Loss of mitochondrial DNA in the yeast cardiolipin synthase crd1 mutant leads to up-regulation of the protein kinase Swe1p that regulates the G2/M transition.

The anionic phospholipid cardiolipin and its precursor phosphatidylglycerol are synthesized and localized in the mitochondrial inner membrane of eukaryotes. They are required for structural integrity and optimal activities of a large number of mitochondrial proteins and complexes. Previous studies showed that loss of anionic phospholipids leads to cell inviability in the absence of mitochondrial DNA. However, the mechanism linking loss of anionic phospholipids to petite lethality was unclear. To elucidate the mechanism, we constructed a crd1Deltarho degrees mutant, which is viable and mimics phenotypes of pgs1Delta in the petite background. We found that loss of cardiolipin in rho degrees cells leads to elevated expression of Swe1p, a morphogenesis checkpoint protein. Moreover, the retrograde pathway is activated in crd1Deltarho degrees cells, most likely due to the exacerbation of mitochondrial dysfunction. Interestingly, the expression of SWE1 is dependent on retrograde regulation as elevated expression of SWE1 is suppressed by deletion of RTG2 or RTG3. Taken together, these findings indicate that activation of the retrograde pathway leads to up-regulation of SWE1 in crd1Deltarho degrees cells. These results suggest that anionic phospholipids are required for processes that are essential for normal cell division in rho degrees cells.

Interaction trap/two-hybrid system to identify interacting proteins.

The yeast two-hybrid method (or interaction trap) is a powerful technique for detecting protein interactions. The procedure is performed using transcriptional activation of a dual reporter system in yeast to identify interactions between a protein of interest (the bait protein) and the candidate proteins for interaction. The method can be used to screen a protein library for interactions with a bait protein or to test for association between proteins that are expected to interact based on prior evidence. Interaction mating facilitates the screening of a library with multiple bait proteins.

Combining multiple positive training sets to generate confidence scores for protein-protein interactions.

MOTIVATION: High-throughput experimental and computational methods are generating a wealth of protein-protein interaction data for a variety of organisms. However, data produced by current state-of-the-art methods include many false positives, which can hinder the analyses needed to derive biological insights. One way to address this problem is to assign confidence scores that reflect the reliability and biological significance of each interaction. Most previously described scoring methods use a set of likely true positives to train a model to score all interactions in a dataset. A single positive training set, however, may be biased and not representative of true interaction space. RESULTS: We demonstrate a method to score protein interactions by utilizing multiple independent sets of training positives to reduce the potential bias inherent in using a single training set. We used a set of benchmark yeast protein interactions to show that our approach outperforms other scoring methods. Our approach can also score interactions across data types, which makes it more widely applicable than many previously proposed methods. We applied the method to protein interaction data from both Drosophila melanogaster and Homo sapiens. Independent evaluations show that the resulting confidence scores accurately reflect the biological significance of the interactions.