Comparative network-based recovery analysis and proteomic profiling of neurological changes in valproic acid-treated mice.

Despite its prominence for characterization of complex mixtures, LC-MS/MS frequently fails to identify many proteins. Network-based analysis methods, based on protein-protein interaction networks (PPINs), biological pathways, and protein complexes, are useful for recovering non-detected proteins, thereby enhancing analytical resolution. However, network-based analysis methods do come in varied flavors for which the respective efficacies are largely unknown. We compare the recovery performance and functional insights from three distinct instances of PPIN-based approaches, viz., Proteomics Expansion Pipeline (PEP), Functional Class Scoring (FCS), and Maxlink, in a test scenario of valproic acid (VPA)-treated mice. We find that the most comprehensive functional insights, as well as best non-detected protein recovery performance, are derived from FCS utilizing real biological complexes. This outstrips other network-based methods such as Maxlink or Proteomics Expansion Pipeline (PEP). From FCS, we identified known biological complexes involved in epigenetic modifications, neuronal system development, and cytoskeletal rearrangements. This is congruent with the observed phenotype where adult mice showed an increase in dendritic branching to allow the rewiring of visual cortical circuitry and an improvement in their visual acuity when tested behaviorally. In addition, PEP also identified a novel complex, comprising YWHAB, NR1, NR2B, ACTB, and TJP1, which is functionally related to the observed phenotype. Although our results suggest different network analysis methods can produce different results, on the whole, the findings are mutually supportive. More critically, the non-overlapping information each provides can provide greater holistic understanding of complex phenotypes.

Enhancing the utility of Proteomics Signature Profiling (PSP) with Pathway Derived Subnets (PDSs), performance analysis and specialised ontologies.

BACKGROUND: Proteomics Signature Profiling (PSP) is a novel hit-rate based method that proved useful in resolving consistency and coverage issues in proteomics. As a follow-up study, several points need to be addressed: 1/ PSP's generalisability to pathways, 2/ understanding the biological interplay between significant complexes and pathway subnets co-located on the same pathways on our liver cancer dataset, 3/ understanding PSP's false positive rate and 4/ demonstrating that PSP works on other suitable proteomics datasets as well as expanding PSP's analytical resolution via the use of specialised ontologies. RESULTS: 1/ PSP performs well with Pathway-Derived Subnets (PDSs). Comparing the performance of PDSs derived from various pathway databases, we find that an integrative approach is best for optimising analytical resolution. Feature selection also confirms that significant PDSs are closely connected to the cancer phenotype.2/ In liver cancer, correlation studies of significant PSP complexes and PDSs co-localised on the same pathways revealed an interesting relationship between the purine metabolism pathway and two other complexes involved in DNA repair. Our work suggests progression to poor stage requires additional mutations that disrupt DNA repair enzymes.3/ False positive analysis reveals that PSP, applied on both complexes and PDSs, is powerful and precise.4/ Via an expert-curated lipid ontology, we uncovered several interesting lipid-associated complexes that could be associated with cancer progression. Of particular interest is the HMGB1-HMGB2-HSC70-ERP60-GAPDH complex which is also involved in DNA repair. We also demonstrated generalisability of PSP using a non-small-cell lung carcinoma data set. CONCLUSIONS: PSP is a powerful and precise technique, capable of identifying biologically coherent features. It works with biological complexes, network-predicted clusters as well as PDSs. Here, an instance of the interplay between significant PDSs and complexes, possibly significantly involved in liver cancer progression but not well understood as yet, is demonstrated. Also demonstrated is the enhancement of PSP's analytical resolution using specialised ontologies.

A network-based maximum link approach towards MS identifies potentially important roles for undetected ARRB1/2 and ACTB in liver cancer progression.

Hepatocellular Carcinoma (HCC) ranks among the deadliest of cancers and has a complex etiology. Proteomics analysis using iTRAQ provides a direct way to analyse perturbations in protein expression during HCC progression from early- to late-stage but suffers from consistency and coverage issues. Appropriate use of network-based analytical methods can help to overcome these issues. We built an integrated and comprehensive Protein-Protein Interaction Network (PPIN) by merging several major databases. Additionally, the network was filtered for GO coherent edges. Significantly differential genes (seeds) were selected from iTRAQ data and mapped onto this network. Undetected proteins linked to seeds (linked proteins) were identified and functionally characterised. The process of network cleaning provides a list of higher quality linked proteins, which are highly enriched for similar biological process gene ontology terms. Linked proteins are also enriched for known cancer genes and are linked to many well-established cancer processes such as apoptosis and immune response. We found that there is an increased propensity for known cancer genes to be found in highly linked proteins. Three highly-linked proteins were identified that may play an important role in driving HCC progression - the G-protein coupled receptor signalling proteins, ARRB1/2 and the structural protein beta-actin, ACTB. Interestingly, both ARRB proteins evaded detection in the iTRAQ screen. ACTB was not detected in the original dataset derived from Mascot but was found to be strongly supported when we re-ran analysis using another protein detection database (Paragon). Identification of linked proteins helps to partially overcome the coverage issue in shotgun proteomics analysis. The set of linked proteins are found to be enriched for cancer-specific processes, and more likely so if they are more highly linked. Additionally, a higher quality linked set is derived if network-cleaning is performed prior. This form of network-based analysis complements the cluster-based approach, and can provide a larger list of proteins on which to perform functional analysis, as well as for biomarker identification.

Proteomics signature profiling (PSP): a novel contextualization approach for cancer proteomics.

Traditional proteomics analysis is plagued by the use of arbitrary thresholds resulting in large loss of information. We propose here a novel method in proteomics that utilizes all detected proteins. We demonstrate its efficacy in a proteomics screen of 5 and 7 liver cancer patients in the moderate and late stage, respectively. Utilizing biological complexes as a cluster vector, and augmenting it with submodules obtained from partitioning an integrated and cleaned protein-protein interaction network, we calculate a Proteomics Signature Profile (PSP) for each patient based on the hit rates of their reported proteins, in the absence of fold change thresholds, against the cluster vector. Using this, we demonstrated that moderate- and late-stage patients segregate with high confidence. We also discovered a moderate-stage patient who displayed a proteomics profile similar to other poor-stage patients. We identified significant clusters using a modified version of the SNet approach. Comparing our results against the Proteomics Expansion Pipeline (PEP) on which the same patient data was analyzed, we found good correlation. Building on this finding, we report significantly more clusters (176 clusters here compared to 70 in PEP), demonstrating the sensitivity of this approach. Gene Ontology (GO) terms analysis also reveals that the significant clusters are functionally congruent with the liver cancer phenotype. PSP is a powerful and sensitive method for analyzing proteomics profiles even when sample sizes are small. It does not rely on the ratio scores but, rather, whether a protein is detected or not. Although consistency of individual proteins between patients is low, we found the reported proteins tend to hit clusters in a meaningful and informative manner. By extracting this information in the form of a Proteomics Signature Profile, we confirm that this information is conserved and can be used for (1) clustering of patient samples, (2) identification of significant clusters based on real biological complexes, and (3) overcoming consistency and coverage issues prevalent in proteomics data sets.

The role of miRNAs in complex formation and control.

UNLABELLED: microRibonucleic acid (miRNAs) are small regulatory molecules that act by mRNA degradation or via translational repression. Although many miRNAs are ubiquitously expressed, a small subset have differential expression patterns that may give rise to tissue-specific complexes. MOTIVATION: This work studies gene targeting patterns amongst miRNAs with differential expression profiles, and links this to control and regulation of protein complexes. RESULTS: We find that, when a pair of miRNAs are not expressed in the same tissues, there is a higher tendency for them to target the direct partners of the same hub proteins. At the same time, they also avoid targeting the same set of hub-spokes. Moreover, the complexes corresponding to these hub-spokes tend to be specific and nonoverlapping. This suggests that the effect of miRNAs on the formation of complexes is specific.

Capturing expert knowledge with argumentation: a case study in bioinformatics.

MOTIVATION: The output of a bioinformatic tool such as BLAST must usually be interpreted by an expert before reliable conclusions can be drawn. This may be based upon the expert's experience, additional data and statistical analysis. Often the process is laborious, goes unrecorded and may be biased. Argumentation is an established technique for reasoning about situations where absolute truth or precise probability is impossible to determine. RESULTS: We demonstrate the application of argumentation to 3D-PSSM, a protein structure prediction tool. The expert's interpretation of results is represented as an argumentation framework. Given a 3D-PSSM result, an automated procedure constructs arguments for and against the conclusion that the result is a good predictor of protein structure. In addition to capturing the unique expertise of the author of 3D-PSSM for distribution to users, an improvement in recall of 5-10 percentage points is achieved. This technique can be applied to a wide range of bioinformatic tools. AVAILABILITY: Example public server and benchmarking data are available at http://www.sbg.bio.ic.ac.uk/~brj03/argumentation/paper/. Source code available on request.

SEAN: SNP prediction and display program utilizing EST sequence clusters.

SEAN is an application that predicts single nucleotide polymorphisms (SNPs) using multiple sequence alignments produced from expressed sequence tag (EST) clusters. The algorithm uses rules of sequence identity and SNP abundance to determine the quality of the prediction. A Java viewer is provided to display the EST alignments and predicted SNPs.

Visualization of microarray results to assist interpretation.

Whole genome microarrays allow assessment of the profile of genes expressed under particular experimental conditions, including external stimuli such as pH or temperature, and internal changes brought about by deleting or over-expressing a gene. Such experiments produce large data sets, for which sophisticated analysis software is available. What is lacking are tools for analysing data sets from different experiments, in order to test and generate hypotheses about the links between regulatory networks. We describe here a method for presenting results from different experiments as a directed graph constructed using an automated graph drawing program xneato, enhanced by a logic program designed to cluster data and aid in the generation of hypotheses about possible gene interactions. A web-based front-end to the system has been constructed to explore and manipulate the graphical displays produced. Results of microarray experiments on Mycobacterium tuberculosis were used to develop and evaluate the visualization tool and initiate the development of an inference system for gene interactions based on such data. The GeneGraph project can be accessed at: zebrafish.doc.ic.ac.uk

GANESH: software for customized annotation of genome regions.

GANESH is a software package designed to support the genetic analysis of regions of human and other genomes. It provides a set of components that may be assembled to construct a self-updating database of DNA sequence, mapping data, and annotations of possible genome features. Once one or more remote sources of data for the target region have been identified, all sequences for that region are downloaded, assimilated, and subjected to a (configurable) set of standard database-searching and genome-analysis packages. The results are stored in compressed form in a relational database, and are updated automatically on a regular schedule so that they are always immediately available in their most up-to-date versions. A Java front-end, executed as a stand alone application or web applet, provides a graphical interface for navigating the database and for viewing the annotations. There are facilities for importing and exporting data in the format of the Distributed Annotation System (DAS), enabling a GANESH database to be used as a component of a DAS configuration. The system has been used to construct databases for about a dozen regions of human chromosomes and for three regions of mouse chromosomes.

An integrated genetic, radiation hybrid, physical and transcription map of a region of distal mouse chromosome 12, including an imprinted locus and the 'Legs at odd angles' (Loa) mutation.

A variety of loci with interesting patterns of regulation such as imprinted expression, and critical functions such as involvement in tumour necrosis factor pathways, map to a distal portion of mouse chromosome 12. This region also contains disease related loci including the 'Legs at odd angles' mutation (Loa) that we are pursuing in a positional cloning project. To further define the region and prepare for comparative sequencing projects, we have produced genetic, radiation hybrid, physical and transcript maps of the region, with probes providing anchors between the maps. We show a summary of 95 markers and 91 genomic clones that has enabled us to identify 18 transcripts including new genes and candidates for Loa which will help in future studies of gene context and regulation.