Markov Clustering Research Articles

Profiling of gene duplication patterns of sequenced teleost genomes: evidence for rapid lineage-specific genome expansion mediated by recent tandem duplications

BackgroundGene duplication has had a major impact on genome evolution. Localized (or tandem) duplication resulting from unequal crossing over and whole genome duplication are believed to be the two dominant mechanisms contributing to vertebrate genome evolution. While much scrutiny has been directed toward discerning patterns indicative of whole-genome duplication events in teleost species, less attention has been paid to the continuous nature of gene duplications and their impact on the size, gene content, functional diversity, and overall architecture of teleost genomes.ResultsHere, using a Markov clustering algorithm directed approach we catalogue and analyze patterns of gene duplication in the four model teleost species with chromosomal coordinates: zebrafish, medaka, stickleback, and Tetraodon. Our analyses based on set size, duplication type, synonymous substitution rate (Ks), and gene ontology emphasize shared and lineage-specific patterns of genome evolution via gene duplication. Most strikingly, our analyses highlight the extraordinary duplication and retention rate of recent duplicates in zebrafish and their likely role in the structural and functional expansion of the zebrafish genome. We find that the zebrafish genome is remarkable in its large number of duplicated genes, small duplicate set size, biased Ks distribution toward minimal mutational divergence, and proportion of tandem and intra-chromosomal duplicates when compared with the other teleost model genomes. The observed gene duplication patterns have played significant roles in shaping the architecture of teleost genomes and appear to have contributed to the recent functional diversification and divergence of important physiological processes in zebrafish.ConclusionsWe have analyzed gene duplication patterns and duplication types among the available teleost genomes and found that a large number of genes were tandemly and intrachromosomally duplicated, suggesting their origin of independent and continuous duplication. This is particularly true for the zebrafish genome. Further analysis of the duplicated gene sets indicated that a significant portion of duplicated genes in the zebrafish genome were of recent, lineage-specific duplication events. Most strikingly, a subset of duplicated genes is enriched among the recently duplicated genes involved in immune or sensory response pathways. Such findings demonstrated the significance of continuous gene duplication as well as that of whole genome duplication in the course of genome evolution.

Coordinating Environmental Genomics and Geochemistry Reveals Metabolic Transitions in a Hot Spring Ecosystem

We have constructed a conceptual model of biogeochemical cycles and metabolic and microbial community shifts within a hot spring ecosystem via coordinated analysis of the “Bison Pool” (BP) Environmental Genome and a complementary contextual geochemical dataset of ∼75 geochemical parameters. 2,321 16S rRNA clones and 470 megabases of environmental sequence data were produced from biofilms at five sites along the outflow of BP, an alkaline hot spring in Sentinel Meadow (Lower Geyser Basin) of Yellowstone National Park. This channel acts as a >22 m gradient of decreasing temperature, increasing dissolved oxygen, and changing availability of biologically important chemical species, such as those containing nitrogen and sulfur. Microbial life at BP transitions from a 92°C chemotrophic streamer biofilm community in the BP source pool to a 56°C phototrophic mat community. We improved automated annotation of the BP environmental genomes using BLAST-based Markov clustering. We have also assigned environmental genome sequences to individual microbial community members by complementing traditional homology-based assignment with nucleotide word-usage algorithms, allowing more than 70% of all reads to be assigned to source organisms. This assignment yields high genome coverage in dominant community members, facilitating reconstruction of nearly complete metabolic profiles and in-depth analysis of the relation between geochemical and metabolic changes along the outflow. We show that changes in environmental conditions and energy availability are associated with dramatic shifts in microbial communities and metabolic function. We have also identified an organism constituting a novel phylum in a metabolic “transition” community, located physically between the chemotroph- and phototroph-dominated sites. The complementary analysis of biogeochemical and environmental genomic data from BP has allowed us to build ecosystem-based conceptual models for this hot spring, reconstructing whole metabolic networks in order to illuminate community roles in shaping and responding to geochemical variability.

GeSICA: Genome segmentation from intra-chromosomal associations

BackgroundVarious aspects of genome organization have been explored based on data from distinct technologies, including histone modification ChIP-Seq, 3C, and its derivatives. Recently developed Hi-C techniques enable the genome wide mapping of DNA interactomes, thereby providing the opportunity to study genome organization in detail, but these methods also pose challenges in methodology development.ResultsWe developed Genome Segmentation from Intra Chromosomal Associations, or GeSICA, to explore genome organization and applied the method to Hi-C data in human GM06990 and K562 cells. GeSICA calculates a simple logged ratio to efficiently segment the human genome into regions with two distinct states that correspond to rich and poor functional element states. Inside the rich regions, Markov Clustering was subsequently applied to segregate the regions into more detailed clusters. The binding sites of the insulator, cohesion, and transcription complexes are enriched in the boundaries between neighboring clusters, indicating that inferred clusters may have fine organizational features.ConclusionsOur study presents a novel analysis method, known as GeSICA, which gives insight into genome organization based on Hi-C data. GeSICA is open source and freely available at: http://web.tongji.edu.cn/~zhanglab/GeSICA/

Fast Parallel Markov Clustering in Bioinformatics Using Massively Parallel Computing on GPU with CUDA and ELLPACK-R Sparse Format

Markov clustering (MCL) is becoming a key algorithm within bioinformatics for determining clusters in networks. However,with increasing vast amount of data on biological networks, performance and scalability issues are becoming a critical limiting factor in applications. Meanwhile, GPU computing, which uses CUDA tool for implementing a massively parallel computing environment in the GPU card, is becoming a very powerful, efficient, and low-cost option to achieve substantial performance gains over CPU approaches. The use of on-chip memory on the GPU is efficiently lowering the latency time, thus, circumventing a major issue in other parallel computing environments, such as MPI. We introduce a very fast Markov clustering algorithm using CUDA (CUDA-MCL) to perform parallel sparse matrix-matrix computations and parallel sparse Markov matrix normalizations, which are at the heart of MCL. We utilized ELLPACK-R sparse format to allow the effective and fine-grain massively parallel processing to cope with the sparse nature of interaction networks data sets in bioinformatics applications. As the results show, CUDA-MCL is significantly faster than the original MCL running on CPU. Thus, large-scale parallel computation on off-the-shelf desktop-machines, that were previously only possible on supercomputing architectures, can significantly change the way bioinformaticians and biologists deal with their data.

High Performance Computational Systems Biology

This paper includes a selection of papers presented at the Second International Workshop on High Performance Computational Systems Biology (HiBi 2010), which was colocated with the joint ICGT/SPIN conference, that was held at the University of Twente, Enschede, The Netherlands, on 27-29 September 2010. The selected papers cover a broad range of bioinformatics topics, including medical imaging, Markov clustering, simulation, model checking, and gene networks.

Massive human co-expression network and its medical applications.

Network-based analysis is indispensable in analyzing high-throughput biological data. Based on the assumption that the variation of gene interactions under given biological conditions could be better interpreted in the context of a large-scale and wide variety of developmental, tissue, and disease conditions, we leverage the large quantity of publicly available transcriptomic data >40,000 HG U133A Affymetrix microarray chips stored in ArrayExpress (http://www.ebi.ac.uk/arrayexpress/) using MetaOmGraph (http://metnet.vrac.iastate.edu/MetNet_MetaOmGraph.htm). From this data, 18,637 chips encompassing over 500 experiments containing high-quality data (18637 Hu-dataset) were used to create a globally stable gene co-expression network (18637 Hu-co-expression-network). Regulons, groups of highly and consistently co-expressed genes, were obtained by partitioning the 18637 Hu-co-expression-network using an Markov clustering algorithm (MCL). The regulons were demonstrated to be statistically significant using a gene ontology (GO) term overrepresentation test combined with evaluation of the effects of gene permutations. The regulons include ca. 12% of human genes, interconnected by 31,471 correlations. All network data and metadata are publically available (http://metnet.vrac.iastate.edu/MetNet_MetaOmGraph.htm). Text mining of these metadata, GO term overrepresentation analysis, and statistical analysis of transcriptomic experiments across multiple environmental, tissue, and disease conditions, has revealed novel fingerprints distinguishing central nervous system (CNS)-related conditions. This study demonstrates the value of mega-scale network-based analysis for biologists to further refine transcriptomic data, derived from a particular condition, to study the global relationships between genes and diseases, and to develop hypotheses that can inform future research.

Information dynamics algorithm for detecting communities in networks

The problem of community detection is relevant in many scientific disciplines, from social science to statistical physics. Given the impact of community detection in many areas, such as psychology and social sciences, we have addressed the issue of modifying existing well performing algorithms by incorporating elements of the domain application fields, i.e. domain-inspired. We have focused on a psychology and social network-inspired approach which may be useful for further strengthening the link between social network studies and mathematics of community detection. Here we introduce a community-detection algorithm derived from the van Dongen’s Markov Cluster algorithm (MCL) method [4] by considering networks’ nodes as agents capable to take decisions. In this framework we have introduced a memory factor to mimic a typical human behavior such as the oblivion effect. The method is based on information diffusion and it includes a non-linear processing phase. We test our method on two classical community benchmark and on computer generated networks with known community structure. Our approach has three important features: the capacity of detecting overlapping communities, the capability of identifying communities from an individual point of view and the fine tuning the community detectability with respect to prior knowledge of the data. Finally we discuss how to use a Shannon entropy measure for parameter estimation in complex networks.

Using a Genetic Algorithm and Markov Clustering on Protein–Protein Interaction Graphs

In this paper, a Genetic Algorithm is applied on the filter of the Enhanced Markov Clustering algorithm to optimize the selection of clusters having a high probability to represent protein complexes. The filter was applied on the results (obtained by experiments made on five different yeast datasets) of three different algorithms known for their efficiency on protein complex detection through protein interaction graphs. The results are compared with three popular clustering algorithms, proving the efficiency of the proposed method according to metrics such as successful prediction rate and geometrical accuracy.

Using hierarchical clustering of secreted protein families to classify and rank candidate effectors of rust fungi.

Rust fungi are obligate biotrophic pathogens that cause considerable damage on crop plants. Puccinia graminis f. sp. tritici, the causal agent of wheat stem rust, and Melampsora larici-populina, the poplar leaf rust pathogen, have strong deleterious impacts on wheat and poplar wood production, respectively. Filamentous pathogens such as rust fungi secrete molecules called disease effectors that act as modulators of host cell physiology and can suppress or trigger host immunity. Current knowledge on effectors from other filamentous plant pathogens can be exploited for the characterisation of effectors in the genome of recently sequenced rust fungi. We designed a comprehensive in silico analysis pipeline to identify the putative effector repertoire from the genome of two plant pathogenic rust fungi. The pipeline is based on the observation that known effector proteins from filamentous pathogens have at least one of the following properties: (i) contain a secretion signal, (ii) are encoded by in planta induced genes, (iii) have similarity to haustorial proteins, (iv) are small and cysteine rich, (v) contain a known effector motif or a nuclear localization signal, (vi) are encoded by genes with long intergenic regions, (vii) contain internal repeats, and (viii) do not contain PFAM domains, except those associated with pathogenicity. We used Markov clustering and hierarchical clustering to classify protein families of rust pathogens and rank them according to their likelihood of being effectors. Using this approach, we identified eight families of candidate effectors that we consider of high value for functional characterization. This study revealed a diverse set of candidate effectors, including families of haustorial expressed secreted proteins and small cysteine-rich proteins. This comprehensive classification of candidate effectors from these devastating rust pathogens is an initial step towards probing plant germplasm for novel resistance components.

Comparison and evaluation of network clustering algorithms applied to genetic interaction networks

The goal of network clustering algorithms detect dense clusters in a network, and provide a first step towards the understanding of large scale biological networks. With numerous recent advances in biotechnologies, large-scale genetic interactions are widely available, but there is a limited understanding of which clustering algorithms may be most effective. In order to address this problem, we conducted a systematic study to compare and evaluate six clustering algorithms in analyzing genetic interaction networks, and investigated influencing factors in choosing algorithms. The algorithms considered in this comparison include hierarchical clustering, topological overlap matrix, bi-clustering, Markov clustering, Bayesian discriminant analysis based community detection, and variational Bayes approach to modularity. Both experimentally identified and synthetically constructed networks were used in this comparison. The accuracy of the algorithms is measured by the Jaccard index in comparing predicted gene modules with benchmark gene sets. The results suggest that the choice differs according to the network topology and evaluation criteria. Hierarchical clustering showed to be best at predicting protein complexes; Bayesian discriminant analysis based community detection proved best under epistatic miniarray profile (EMAP) datasets; the variational Bayes approach to modularity was noticeably better than the other algorithms in the genome-scale networks.

Comparison and evaluation of network clustering algorithms applied to genetic interaction networks

The goal of network clustering algorithms detect dense clusters in a network, and provide a first step towards the understanding of large scale biological networks. With numerous recent advances in biotechnologies, large-scale genetic interactions are widely available, but there is a limited understanding of which clustering algorithms may be most effective. In order to address this problem, we conducted a systematic study to compare and evaluate six clustering algorithms in analyzing genetic interaction networks, and investigated influencing factors in choosing algorithms. The algorithms considered in this comparison include hierarchical clustering, topological overlap matrix, bi-clustering, Markov clustering, Bayesian discriminant analysis based community detection, and variational Bayes approach to modularity. Both experimentally identified and synthetically constructed networks were used in this comparison. The accuracy of the algorithms is measured by the Jaccard index in comparing predicted gene modules with benchmark gene sets. The results suggest that the choice differs according to the network topology and evaluation criteria. Hierarchical clustering showed to be best at predicting protein complexes; Bayesian discriminant analysis based community detection proved best under epistatic miniarray profile (EMAP) datasets; the variational Bayes approach to modularity was noticeably better than the other algorithms in the genome-scale networks.

Improving Web Page Prediction Using Default Rule Selection

Mining user patterns of web log files can provide significant and useful informative knowledge. A large amount of research has been done in trying to predict correctly the pages a user will most likely request next. Markov models are the most commonly used approaches for this type of web access prediction. Web page prediction requires the development of models that can predict a user’s next access to a web server. Many researchers have proposed a novel approach that integrates Markov models, association rules and clustering in web site access predictability. The low order Markov models provide higher coverage, but these are couched in ambiguous rules. In this paper, we introduce the use of default rule in resolving web access ambiguous predictions. This method could provide better prediction than using the individual traditional models. The results have shown that the default rule increases the accuracy and model-accuracy of web page access predictions. It also applies to association rules and the other combined models.

Video Behaviour Mining Using a Dynamic Topic Model

This paper addresses the problem of fully automated mining of public space video data, a highly desirable capability under contemporary commercial and security considerations. This task is especially challenging due to the complexity of the object behaviors to be profiled, the difficulty of analysis under the visual occlusions and ambiguities common in public space video, and the computational challenge of doing so in real-time. We address these issues by introducing a new dynamic topic model, termed a Markov Clustering Topic Model (MCTM). The MCTM builds on existing dynamic Bayesian network models and Bayesian topic models, and overcomes their drawbacks on sensitivity, robustness and efficiency. Specifically, our model profiles complex dynamic scenes by robustly clustering visual events into activities and these activities into global behaviours with temporal dynamics. A Gibbs sampler is derived for offline learning with unlabeled training data and a new approximation to online Bayesian inference is formulated to enable dynamic scene understanding and behaviour mining in new video data online in real-time. The strength of this model is demonstrated by unsupervised learning of dynamic scene models for four complex and crowded public scenes, and successful mining of behaviors and detection of salient events in each.

Whole-Exome Sequencing of 8 Atypical Chronic Myeloid Leukaemia Patients,

Abstract 3853 INTRODUCTIONAccording to the WHO-2008 classification, atypical chronic myeloid leukemia (aCML) is a myelodysplastic/myeloproliferative disorder clinically resembling the Philadelphia positive CML but lacking the BCR-ABL fusion. In aCML, the molecular lesions underlying the onset of the disease are unknown. To investigate the somatic events occurring in the genome of the aCML leukemic cells, we carried out whole-exome high-throughput sequencing analyses of 8 aCML patients. The results are described here. METHODSPeripheral blood (PB) or bone marrow cells were obtained after informed consent at diagnosis, before any therapy. Myeloid cells, evaluated by FACS, constituted more than 80% of total cells. Lymphocytes were obtained from PB samples, after culture with PHA/IL2 for 2–3 weeks. The exon-capture protocol was performed on myeloid leukemic cells and normal lymphocytes from the same patients using the Illumina TruSeq Exome Enrichment Kit. The enriched DNA was sequenced with a Genome Analyzer IIx (Illumina), using a 60 bases paired-end protocol and the TruSeq chemistry. On average, 10.5 Gigabases per exome were generated. The bioinformatic analysis was performed using the Galaxy framework (http://main.g2.bx.psu.edu/); the cross-match between leukemic and normal exomes was performed with dedicated in-house C# software. RESULTSThe percentage of reads matching the reference human genome was over 90%, with a mean exon coverage of >70-fold and a percentage of exons with a mean coverage ≥ 20x of > 90% for both the leukemic sample and the control. The percentage of nucleotides targeting exonic regions or exonic regions plus 100bp was 48% and 68%, respectively, with an overall 28-fold enrichment for exonic vs. non-exonic regions. The comparison between the leukemic and the control datasets led to the identification of 63 single nucleotide somatic mutations with a relative mutation coverage of > 35%, corresponding to their heterozygous presence in >88% of cells. In total, 46 mutations were transitions and 17 transversions (transition/transversion ratio of 2.7), with the C:G->T:A event occurring at highest frequency (35/63; 55.6%). In 19/35 (54.3%) of the cases, the C:G->T:A transition occurred in the context of a CpG site. Among the 63 variants, 32.8% ranked more than 1.0 and 21.3% more than 2.0 in the GeneRanker cancer scoring system (http://cbio.mskcc.org/tcga-generanker/). Characterization of the top scoring biological functions (Ingenuity Pathway Analysis software) revealed a strong association with the core functional concepts of ‘Cancer/Leukaemia’ (p = 3.74 * 10−6) and ‘Myeloproliferative disorder’ (p = 9.91 * 10−6), with a total of 23 genes in the Cancer and 18 in the Leukaemia functional annotation. The top scoring cellular functions were connected with the core concepts of ‘Cellular growth and proliferation’ (p = 1.03 * 10−3; 17 genes) and ‘DNA replication, recombination and repair’ (5 genes). The latter comprised the functional annotations: ‘Chromatin formation’, ‘Chromatin remodelling’ and ‘Formation of chromosome components’ (p = 6.76 * 10−3, 1.96 * 10−2 and 7.76 * 10−4, respectively). A Markov Cluster Algorithm analysis (String, http://string-db.org), confirmed the presence of the same two main mutational clusters (Fig. 1a,b). The individual mutations identified will be presented at the meeting, including two recurrent mutations. Taken globally, these data suggest that somatic mutations targeting a specific cellular proliferation pathway and a chromatin remodelling protein network (Fig. 1a,b) may play a critical role in the onset/clonal evolution of aCML. This information may prove useful in the next future in order to develop evidence-based aCML treatment protocols. [Display omitted] Disclosures:No relevant conflicts of interest to declare.

A binary matrix factorization algorithm for protein complex prediction

BackgroundIdentifying biologically relevant protein complexes from a large protein-protein interaction (PPI) network, is essential to understand the organization of biological systems. However, high-throughput experimental techniques that can produce a large amount of PPIs are known to yield non-negligible rates of false-positives and false-negatives, making the protein complexes difficult to be identified.ResultsWe propose a binary matrix factorization (BMF) algorithm under the Bayesian Ying-Yang (BYY) harmony learning, to detect protein complexes by clustering the proteins which share similar interactions through factorizing the binary adjacent matrix of a PPI network. The proposed BYY-BMF algorithm automatically determines the cluster number while this number is pre-given for most existing BMF algorithms. Also, BYY-BMF’s clustering results does not depend on any parameters or thresholds, unlike the Markov Cluster Algorithm (MCL) that relies on a so-called inflation parameter. On synthetic PPI networks, the predictions evaluated by the known annotated complexes indicate that BYY-BMF is more robust than MCL for most cases. On real PPI networks from the MIPS and DIP databases, BYY-BMF obtains a better balanced prediction accuracies than MCL and a spectral analysis method, while MCL has its own advantages, e.g., with good separation values.

QAARM: quasi-anharmonic autoregressive model reveals molecular recognition pathways in ubiquitin.

Motivation: Molecular dynamics (MD) simulations have dramatically improved the atomistic understanding of protein motions, energetics and function. These growing datasets have necessitated a corresponding emphasis on trajectory analysis methods for characterizing simulation data, particularly since functional protein motions and transitions are often rare and/or intricate events. Observing that such events give rise to long-tailed spatial distributions, we recently developed a higher-order statistics based dimensionality reduction method, called quasi-anharmonic analysis (QAA), for identifying biophysically-relevant reaction coordinates and substates within MD simulations. Further characterization of conformation space should consider the temporal dynamics specific to each identified substate.Results: Our model uses hierarchical clustering to learn energetically coherent substates and dynamic modes of motion from a 0.5 μs ubiqutin simulation. Autoregressive (AR) modeling within and between states enables a compact and generative description of the conformational landscape as it relates to functional transitions between binding poses. Lacking a predictive component, QAA is extended here within a general AR model appreciative of the trajectory's temporal dependencies and the specific, local dynamics accessible to a protein within identified energy wells. These metastable states and their transition rates are extracted within a QAA-derived subspace using hierarchical Markov clustering to provide parameter sets for the second-order AR model. We show the learned model can be extrapolated to synthesize trajectories of arbitrary length.Contact: ramanathana@ornl.gov; chakracs@pitt.edu

An Effective Comparison of Graph Clustering Algorithms via Random Graphs

graph clustering algorithms have been proposed in recent past researches, each algorithm having its own advantages and drawbacks. All these algorithms rely on a very different approach so it's really hard to say that which one is the most efficient and optimal if we talk in the sense of performance. It is really hard to decide that which algorithm is beneficial in case of highly complex networks like PPI networks which consist of thousands of nodes. The paper proposes an effective data comparison of RNSC (Restricted Neighbourhood Search Clustering) and MCL (Markov Clustering) algorithms based on Erdos-Renyi and Power-Law Distribution graphs. The basic parameters used for comparison are Edge Density, Run Time, Number of Nodes, Cluster Size and Singleton Cluster. Our approach is an effective one because firstly we have used two types of graph generators, Erdos-Renyi and Scaled-Free for generation of input graphs which are very much closer to the real input graphs and secondly we have generated input graphs having more than 1000 nodes, so in our approach we have used both the algorithms for clustering highly complex input graphs just like PPI networks. For comparison and analysis purpose we have collected data sets and generated some graphs based on these parameters. The proposed approach depicts which algorithm is best to be used for clustering such complex graphs and also some fields for extension if possible in both them. All graphs used in this thesis are unweighted and undirected.

Assessing the functional coherence of modules found in multiple-evidence networks from Arabidopsis

BackgroundCombining multiple evidence-types from different information sources has the potential to reveal new relationships in biological systems. The integrated information can be represented as a relationship network, and clustering the network can suggest possible functional modules. The value of such modules for gaining insight into the underlying biological processes depends on their functional coherence. The challenges that we wish to address are to define and quantify the functional coherence of modules in relationship networks, so that they can be used to infer function of as yet unannotated proteins, to discover previously unknown roles of proteins in diseases as well as for better understanding of the regulation and interrelationship between different elements of complex biological systems.ResultsWe have defined the functional coherence of modules with respect to the Gene Ontology (GO) by considering two complementary aspects: (i) the fragmentation of the GO functional categories into the different modules and (ii) the most representative functions of the modules. We have proposed a set of metrics to evaluate these two aspects and demonstrated their utility in Arabidopsis thaliana. We selected 2355 proteins for which experimentally established protein-protein interaction (PPI) data were available. From these we have constructed five relationship networks, four based on single types of data: PPI, co-expression, co-occurrence of protein names in scientific literature abstracts and sequence similarity and a fifth one combining these four evidence types. The ability of these networks to suggest biologically meaningful grouping of proteins was explored by applying Markov clustering and then by measuring the functional coherence of the clusters.ConclusionsRelationship networks integrating multiple evidence-types are biologically informative and allow more proteins to be assigned to a putative functional module. Using additional evidence types concentrates the functional annotations in a smaller number of modules without unduly compromising their consistency. These results indicate that integration of more data sources improves the ability to uncover functional association between proteins, both by allowing more proteins to be linked and producing a network where modular structure more closely reflects the hierarchy in the gene ontology.

ReMark: an automatic program for clustering orthologs flexibly combining a Recursive and a Markov clustering algorithms

ReMark is a fully automatic tool for clustering orthologs by combining a Recursive and a Markov clustering (MCL) algorithms. The ReMark detects and recursively clusters ortholog pairs through reciprocal BLAST best hits between multiple genomes running software program (RecursiveClustering.java) in the first step. Then, it employs MCL algorithm to compute the clusters (score matrices generated from the previous step) and refines the clusters by adjusting an inflation factor running software program (MarkovClustering.java). This method has two key features. One utilizes, to get more reliable results, the diagonal scores in the matrix of the initial ortholog clusters. Another clusters orthologs flexibly through being controlled naturally by MCL with a selected inflation factor. Users can therefore select the fitting state of orthologous protein clusters by regulating the inflation factor according to their research interests. Source code for the orthologous protein clustering software is freely available for non-commercial use at http://dasan.sejong.ac.kr/~wikim/notice.html, implemented in Java 1.6 and supported on Windows and Linux.

GRAONTO: A graph-based approach for automatic construction of domain ontology

Extracting domain knowledge and taking its full advantage has been an important way to reducing costs and accelerating processes in domain-related applications. Domain ontology, providing a common and unambiguous understanding of a domain for both the users and the system to communicate with each other via a set of representational primitives, has been proposed as an important and natural approach to represent domain knowledge. Most domain knowledge about domain entities with their properties and relationships is embodied in document collections. Thus, extracting ontologies from these documents is an important means of ontology construction. In this paper, a graph-based approach for automatic construction of domain ontology from domain corpus, named GRAONTO, has been proposed. First, each document in the collection is represented by a graph. After the generation of document graphs, random walk term weighting is employed to estimate the relevance of the information of a term to the corpus from both local and global perspectives. Next, the MCL (Markov Clustering) algorithm is used to disambiguate terms with different meanings and group similar terms to produce concepts. Next, an improved gSpan algorithm constrained by both vertices and informativeness is exploited to find arbitrary latent relations among these concepts. Finally, the domain ontology is output in the OWL format. For ontology evaluation purposes, a method for adaptive adjustment of concepts and relations with respect to its practical effectiveness is conceived. Evaluation experiments show that GRAONTO is a promising approach for domain ontology construction.