From knockouts to networks: establishing direct cause-effect relationships through graph analysis

Background

Reverse-engineering gene networks from expression profiles is a difficult problem for which a multitude of techniques have been developed over the last decade. The yearly organized DREAM challenges allow for a fair evaluation and unbiased comparison of these methods.

Results

We propose an inference algorithm that combines confidence matrices, computed as the standard scores from single-gene knockout data, with the down-ranking of feed-forward edges. Substantial improvements on the predictions can be obtained after the execution of this second step.

Conclusions

Our algorithm was awarded the best overall performance at the DREAM4 In Silico 100-gene network sub-challenge, proving to be effective in inferring medium-size gene regulatory networks. This success demonstrates once again the decisive importance of gene expression data obtained after systematic gene perturbations and highlights the usefulness of graph analysis to increase the reliability of inference.     

Dataset size and composition impact the reliability of performance benchmarks for peptide-MHC binding predictions

Background

It is important to accurately determine the performance of peptide:MHC binding predictions, as this enables users to compare and choose between different prediction methods and provides estimates of the expected error rate. Two common approaches to determine prediction performance are cross-validation, in which all available data are iteratively split into training and testing data, and the use of blind sets generated separately from the data used to construct the predictive method. In the present study, we have compared cross-validated prediction performances generated on our last benchmark dataset from 2009 with prediction performances generated on data subsequently added to the Immune Epitope Database (IEDB) which served as a blind set.

Results

We found that cross-validated performances systematically overestimated performance on the blind set. This was found not to be due to the presence of similar peptides in the cross-validation dataset. Rather, we found that small size and low sequence/affinity diversity of either training or blind datasets were associated with large differences in cross-validated vs. blind prediction performances. We use these findings to derive quantitative rules of how large and diverse datasets need to be to provide generalizable performance estimates.

Conclusion

It has long been known that cross-validated prediction performance estimates often overestimate performance on independently generated blind set data. We here identify and quantify the specific factors contributing to this effect for MHC-I binding predictions. An increasing number of peptides for which MHC binding affinities are measured experimentally have been selected based on binding predictions and thus are less diverse than historic datasets sampling the entire sequence and affinity space, making them more difficult benchmark data sets. This has to be taken into account when comparing performance metrics between different benchmarks, and when deriving error estimates for predictions based on benchmark performance.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2105-15-241) contains supplementary material, which is available to authorized users.     

IPred - integrating ab initio and evidence based gene predictions to improve prediction accuracy

Background

Gene prediction is a challenging but crucial part in most genome analysis pipelines. Various methods have evolved that predict genes ab initio on reference sequences or evidence based with the help of additional information, such as RNA-Seq reads or EST libraries. However, none of these strategies is bias-free and one method alone does not necessarily provide a complete set of accurate predictions.

Results

We present IPred (Integrative gene Prediction), a method to integrate ab initio and evidence based gene identifications to complement the advantages of different prediction strategies. IPred builds on the output of gene finders and generates a new combined set of gene identifications, representing the integrated evidence of the single method predictions.

Conclusion

We evaluate IPred in simulations and real data experiments on Escherichia Coli and human data. We show that IPred improves the prediction accuracy in comparison to single method predictions and to existing methods for prediction combination.

Electronic supplementary material

The online version of this article (doi:10.1186/s12864-015-1315-9) contains supplementary material, which is available to authorized users.     

A Quantitative Image Cytometry Technique for Time Series or Population Analyses of Signaling Networks

Background

Modeling of cellular functions on the basis of experimental observation is increasingly common in the field of cellular signaling. However, such modeling requires a large amount of quantitative data of signaling events with high spatio-temporal resolution. A novel technique which allows us to obtain such data is needed for systems biology of cellular signaling.

Methodology/Principal Findings

We developed a fully automatable assay technique, termed quantitative image cytometry (QIC), which integrates a quantitative immunostaining technique and a high precision image-processing algorithm for cell identification. With the aid of an automated sample preparation system, this device can quantify protein expression, phosphorylation and localization with subcellular resolution at one-minute intervals. The signaling activities quantified by the assay system showed good correlation with, as well as comparable reproducibility to, western blot analysis. Taking advantage of the high spatio-temporal resolution, we investigated the signaling dynamics of the ERK pathway in PC12 cells.

Conclusions/Significance

The QIC technique appears as a highly quantitative and versatile technique, which can be a convenient replacement for the most conventional techniques including western blot, flow cytometry and live cell imaging. Thus, the QIC technique can be a powerful tool for investigating the systems biology of cellular signaling.     

MOSBIE: a tool for comparison and analysis of rule-based biochemical models

Background

Mechanistic models that describe the dynamical behaviors of biochemical systems are common in computational systems biology, especially in the realm of cellular signaling. The development of families of such models, either by a single research group or by different groups working within the same area, presents significant challenges that range from identifying structural similarities and differences between models to understanding how these differences affect system dynamics.

Results

We present the development and features of an interactive model exploration system, MOSBIE, which provides utilities for identifying similarities and differences between models within a family. Models are clustered using a custom similarity metric, and a visual interface is provided that allows a researcher to interactively compare the structures of pairs of models as well as view simulation results.

Conclusions

We illustrate the usefulness of MOSBIE via two case studies in the cell signaling domain. We also present feedback provided by domain experts and discuss the benefits, as well as the limitations, of the approach.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2105-15-316) contains supplementary material, which is available to authorized users.     

Epitopemap: a web application for integrated whole proteome epitope prediction

Background

Predictions of MHC binding affinity are commonly used in immunoinformatics for T cell epitope prediction. There are multiple available methods, some of which provide web access. However there is currently no convenient way to access the results from multiple methods at the same time or to execute predictions for an entire proteome at once.

Results

We designed a web application that allows integration of multiple epitope prediction methods for any number of proteins in a genome. The tool is a front-end for various freely available methods. Features include visualisation of results from multiple predictors within proteins in one plot, genome-wide analysis and estimates of epitope conservation.

Conclusions

We present a self contained web application, Epitopemap, for calculating and viewing epitope predictions with multiple methods. The tool is easy to use and will assist in computational screening of viral or bacterial genomes.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-015-0659-0) contains supplementary material, which is available to authorized users.     

Layered Signaling Regulatory Networks Analysis of Gene Expression Involved in Malignant Tumorigenesis of Non-Resolving Ulcerative Colitis via Integration of Cross-Study Microarray Profiles

Background

Ulcerative colitis (UC) was the most frequently diagnosed inflammatory bowel disease (IBD) and closely linked to colorectal carcinogenesis. By far, the underlying mechanisms associated with the disease are still unclear. With the increasing accumulation of microarray gene expression profiles, it is profitable to gain a systematic perspective based on gene regulatory networks to better elucidate the roles of genes associated with disorders. However, a major challenge for microarray data analysis is the integration of multiple-studies generated by different groups.

Methodology/Principal Findings

In this study, firstly, we modeled a signaling regulatory network associated with colorectal cancer (CRC) initiation via integration of cross-study microarray expression data sets using Empirical Bayes (EB) algorithm. Secondly, a manually curated human cancer signaling map was established via comprehensive retrieval of the publicly available repositories. Finally, the co-differently-expressed genes were manually curated to portray the layered signaling regulatory networks.

Results

Overall, the remodeled signaling regulatory networks were separated into four major layers including extracellular, membrane, cytoplasm and nucleus, which led to the identification of five core biological processes and four signaling pathways associated with colorectal carcinogenesis. As a result, our biological interpretation highlighted the importance of EGF/EGFR signaling pathway, EPO signaling pathway, T cell signal transduction and members of the BCR signaling pathway, which were responsible for the malignant transition of CRC from the benign UC to the aggressive one.

Conclusions

The present study illustrated a standardized normalization approach for cross-study microarray expression data sets. Our model for signaling networks construction was based on the experimentally-supported interaction and microarray co-expression modeling. Pathway-based signaling regulatory networks analysis sketched a directive insight into colorectal carcinogenesis, which was of significant importance to monitor disease progression and improve therapeutic interventions.     

Proteomics-Based Systems Biology Modeling of Bovine Germinal Vesicle Stage Oocyte and Cumulus Cell Interaction

Background

Oocytes are the female gametes which establish the program of life after fertilization. Interactions between oocyte and the surrounding cumulus cells at germinal vesicle (GV) stage are considered essential for proper maturation or ‘programming’ of oocytes, which is crucial for normal fertilization and embryonic development. However, despite its importance, little is known about the molecular events and pathways involved in this bidirectional communication.

Methodology/Principal Findings

We used differential detergent fractionation multidimensional protein identification technology (DDF-Mud PIT) on bovine GV oocyte and cumulus cells and identified 811 and 1247 proteins in GV oocyte and cumulus cells, respectively; 371 proteins were significantly differentially expressed between each cell type. Systems biology modeling, which included Gene Ontology (GO) and canonical genetic pathway analysis, showed that cumulus cells have higher expression of proteins involved in cell communication, generation of precursor metabolites and energy, as well as transport than GV oocytes. Our data also suggests a hypothesis that oocytes may depend on the presence of cumulus cells to generate specific cellular signals to coordinate their growth and maturation.

Conclusions/Significance

Systems biology modeling of bovine oocytes and cumulus cells in the context of GO and protein interaction networks identified the signaling pathways associated with the proteins involved in cell-to-cell signaling biological process that may have implications in oocyte competence and maturation. This first comprehensive systems biology modeling of bovine oocytes and cumulus cell proteomes not only provides a foundation for signaling and cell physiology at the GV stage of oocyte development, but are also valuable for comparative studies of other stages of oocyte development at the molecular level.     

Modeling the Effects of Light and Sucrose on In Vitro Propagated Plants: A Multiscale System Analysis Using Artificial Intelligence Technology

Background

Plant acclimation is a highly complex process, which cannot be fully understood by analysis at any one specific level (i.e. subcellular, cellular or whole plant scale). Various soft-computing techniques, such as neural networks or fuzzy logic, were designed to analyze complex multivariate data sets and might be used to model large such multiscale data sets in plant biology.

Methodology and Principal Findings

In this study we assessed the effectiveness of applying neuro-fuzzy logic to modeling the effects of light intensities and sucrose content/concentration in the in vitro culture of kiwifruit on plant acclimation, by modeling multivariate data from 14 parameters at different biological scales of organization. The model provides insights through application of 14 sets of straightforward rules and indicates that plants with lower stomatal aperture areas and higher photoinhibition and photoprotective status score best for acclimation. The model suggests the best condition for obtaining higher quality acclimatized plantlets is the combination of 2.3% sucrose and photonflux of 122–130 µmol m⁻² s⁻¹.

Conclusions

Our results demonstrate that artificial intelligence models are not only successful in identifying complex non-linear interactions among variables, by integrating large-scale data sets from different levels of biological organization in a holistic plant systems-biology approach, but can also be used successfully for inferring new results without further experimental work.     

Effects of clinically relevant MPL mutations in the transmembrane domain revealed at the atomic level through computational modeling

Background

Mutations in the thrombopoietin receptor (MPL) may activate relevant pathways and lead to chronic myeloproliferative neoplasms (MPNs). The mechanisms of MPL activation remain elusive because of a lack of experimental structures. Modern computational biology techniques were utilized to explore the mechanisms of MPL protein activation due to various mutations.

Results

Transmembrane (TM) domain predictions, homology modeling, ab initio protein structure prediction, and molecular dynamics (MD) simulations were used to build structural dynamic models of wild-type and four clinically observed mutants of MPL. The simulation results suggest that S505 and W515 are important in keeping the TM domain in its correct position within the membrane. Mutations at either of these two positions cause movement of the TM domain, altering the conformation of the nearby intracellular domain in unexpected ways, and may cause the unwanted constitutive activation of MPL''s kinase partner, JAK2.

Conclusions

Our findings represent the first full-scale molecular dynamics simulations of the wild-type and clinically observed mutants of the MPL protein, a critical element of the MPL-JAK2-STAT signaling pathway. In contrast to usual explanations for the activation mechanism that are based on the relative translational movement between rigid domains of MPL, our results suggest that mutations within the TM region could result in conformational changes including tilt and rotation (azimuthal) angles along the membrane axis. Such changes may significantly alter the conformation of the adjacent and intrinsically flexible intracellular domain. Hence, caution should be exercised when interpreting experimental evidence based on rigid models of cytokine receptors or similar systems.     

Genetic architecture is more complex for resistance to Septoria tritici blotch than to Fusarium head blight in Central European winter wheat

Background

Fusarium head blight (FHB) and Septoria tritici blotch (STB) severely impair wheat production. With the aim to further elucidate the genetic architecture underlying FHB and STB resistance, we phenotyped 1604 European wheat hybrids and their 135 parental lines for FHB and STB disease severities and determined genotypes at 17,372 single-nucleotide polymorphic loci.

Results

Cross-validated association mapping revealed the absence of large effect QTL for both traits. Genomic selection showed a three times higher prediction accuracy for FHB than STB disease severity for test sets largely unrelated to the training sets.

Conclusions

Our findings suggest that the genetic architecture is less complex and, hence, can be more properly tackled to perform accurate prediction for FHB than STB disease severity. Consequently, FHB disease severity is an interesting model trait to fine-tune genomic selection models exploiting beyond relatedness also knowledge of the genetic architecture.

Electronic supplementary material

The online version of this article (doi:10.1186/s12864-015-1628-8) contains supplementary material, which is available to authorized users.     

CNNcon: Improved Protein Contact Maps Prediction Using Cascaded Neural Networks

Backgrounds

Despite continuing progress in X-ray crystallography and high-field NMR spectroscopy for determination of three-dimensional protein structures, the number of unsolved and newly discovered sequences grows much faster than that of determined structures. Protein modeling methods can possibly bridge this huge sequence-structure gap with the development of computational science. A grand challenging problem is to predict three-dimensional protein structure from its primary structure (residues sequence) alone. However, predicting residue contact maps is a crucial and promising intermediate step towards final three-dimensional structure prediction. Better predictions of local and non-local contacts between residues can transform protein sequence alignment to structure alignment, which can finally improve template based three-dimensional protein structure predictors greatly.

Methods

CNNcon, an improved multiple neural networks based contact map predictor using six sub-networks and one final cascade-network, was developed in this paper. Both the sub-networks and the final cascade-network were trained and tested with their corresponding data sets. While for testing, the target protein was first coded and then input to its corresponding sub-networks for prediction. After that, the intermediate results were input to the cascade-network to finish the final prediction.

Results

The CNNcon can accurately predict 58.86% in average of contacts at a distance cutoff of 8 Å for proteins with lengths ranging from 51 to 450. The comparison results show that the present method performs better than the compared state-of-the-art predictors. Particularly, the prediction accuracy keeps steady with the increase of protein sequence length. It indicates that the CNNcon overcomes the thin density problem, with which other current predictors have trouble. This advantage makes the method valuable to the prediction of long length proteins. As a result, the effective prediction of long length proteins could be possible by the CNNcon.     

Reducing false-positive prediction of minimotifs with a genetic interaction filter

Background

Minimotifs are short contiguous peptide sequences in proteins that have known functions. At its simplest level, the minimotif sequence is present in a source protein and has an activity relationship with a target, most of which are proteins. While many scientists routinely investigate new minimotif functions in proteins, the major web-based discovery tools have a high rate of false-positive prediction. Any new approach that reduces false-positives will be of great help to biologists.

Methods and Findings

We have built three filters that use genetic interactions to reduce false-positive minimotif predictions. The basic filter identifies those minimotifs where the source/target protein pairs have a known genetic interaction. The HomoloGene genetic interaction filter extends these predictions to predicted genetic interactions of orthologous proteins and the node-based filter identifies those minimotifs where proteins that have a genetic interaction with the source or target have a genetic interaction. Each filter was evaluated with a test data set containing thousands of true and false-positives. Based on sensitivity and selectivity performance metrics, the basic filter had the best discrimination for true positives, whereas the node-based filter had the highest sensitivity. We have implemented these genetic interaction filters on the Minimotif Miner 2.3 website. The genetic interaction filter is particularly useful for improving predictions of posttranslational modifications such as phosphorylation and proteolytic cleavage sites.

Conclusions

Genetic interaction data sets can be used to reduce false-positive minimotif predictions. Minimotif prediction in known genetic interactions can help to refine the mechanisms behind the functional connection between genes revealed by genetic experimentation and screens.     

Genomic relationships based on X chromosome markers and accuracy of genomic predictions with and without X chromosome markers

Background

Although the X chromosome is the second largest bovine chromosome, markers on the X chromosome are not used for genomic prediction in some countries and populations. In this study, we presented a method for computing genomic relationships using X chromosome markers, investigated the accuracy of imputation from a low density (7K) to the 54K SNP (single nucleotide polymorphism) panel, and compared the accuracy of genomic prediction with and without using X chromosome markers.

Methods

The impact of considering X chromosome markers on prediction accuracy was assessed using data from Nordic Holstein bulls and different sets of SNPs: (a) the 54K SNPs for reference and test animals, (b) SNPs imputed from the 7K to the 54K SNP panel for test animals, (c) SNPs imputed from the 7K to the 54K panel for half of the reference animals, and (d) the 7K SNP panel for all animals. Beagle and Findhap were used for imputation. GBLUP (genomic best linear unbiased prediction) models with or without X chromosome markers and with or without a residual polygenic effect were used to predict genomic breeding values for 15 traits.

Results

Averaged over the two imputation datasets, correlation coefficients between imputed and true genotypes for autosomal markers, pseudo-autosomal markers, and X-specific markers were 0.971, 0.831 and 0.935 when using Findhap, and 0.983, 0.856 and 0.937 when using Beagle. Estimated reliabilities of genomic predictions based on the imputed datasets using Findhap or Beagle were very close to those using the real 54K data. Genomic prediction using all markers gave slightly higher reliabilities than predictions without X chromosome markers. Based on our data which included only bulls, using a G matrix that accounted for sex-linked relationships did not improve prediction, compared with a G matrix that did not account for sex-linked relationships. A model that included a polygenic effect did not recover the loss of prediction accuracy from exclusion of X chromosome markers.

Conclusions

The results from this study suggest that markers on the X chromosome contribute to accuracy of genomic predictions and should be used for routine genomic evaluation.     

A Digital Network Approach to Infer Sex Behavior in Emerging HIV Epidemics

Purpose

Improve the ability to infer sex behaviors more accurately using network data.

Methods

A hybrid network analytic approach was utilized to integrate: (1) the plurality of reports from others tied to individual(s) of interest; and (2) structural features of the network generated from those ties. Network data was generated from digitally extracted cell-phone contact lists of a purposeful sample of 241 high-risk men in India. These data were integrated with interview responses to describe the corresponding individuals in the contact lists and the ties between them. HIV serostatus was collected for each respondent and served as an internal validation of the model’s predictions of sex behavior.

Results

We found that network-based model predictions of sex behavior and self-reported sex behavior had limited correlation (54% agreement). Additionally, when respondent sex behaviors were re-classified to network model predictions from self-reported data, there was a 30.7% decrease in HIV seroprevalence among groups of men with lower risk behavior, which is consistent with HIV transmission biology.

Conclusion

Combining the relative completeness and objectivity of digital network data with the substantive details of classical interview and HIV biomarker data permitted new analyses and insights into the accuracy of self-reported sex behavior.     

Reduced Context Effects on Retrieval in First-Episode Schizophrenia

Background

A recent modeling study by the authors predicted that contextual information is poorly integrated into episodic representations in schizophrenia, and that this is a main cause of the retrieval deficits seen in schizophrenia.

Methodology/Principal Findings

We have tested this prediction in patients with first-episode schizophrenia and matched controls. The benefit from contextual cues in retrieval was strongly reduced in patients. On the other hand, retrieval based on item cues was spared.

Conclusions/Significance

These results suggest that reduced integration of context information into episodic representations is a core deficit in schizophrenia and one of the main causes of episodic memory impairment.