Binary tree-structured vector quantization approach to clustering and visualizing microarray data

MOTIVATION: With the increasing number of gene expression databases, the need for more powerful analysis and visualization tools is growing. Many techniques have successfully been applied to unravel latent similarities among genes and/or experiments. Most of the current systems for microarray data analysis use statistical methods, hierarchical clustering, self-organizing maps, support vector machines, or k-means clustering to organize genes or experiments into 'meaningful' groups. Without prior explicit bias almost all of these clustering methods applied to gene expression data not only produce different results, but may also produce clusters with little or no biological relevance. Of these methods, agglomerative hierarchical clustering has been the most widely applied, although many limitations have been identified. RESULTS: Starting with a systematic comparison of the underlying theories behind clustering approaches, we have devised a technique that combines tree-structured vector quantization and partitive k-means clustering (BTSVQ). This hybrid technique has revealed clinically relevant clusters in three large publicly available data sets. In contrast to existing systems, our approach is less sensitive to data preprocessing and data normalization. In addition, the clustering results produced by the technique have strong similarities to those of self-organizing maps (SOMs). We discuss the advantages and the mathematical reasoning behind our approach.     

A new algorithm for comparing and visualizing relationships between hierarchical and flat gene expression data clusterings

MOTIVATION: Clustering is one of the most widely used methods in unsupervised gene expression data analysis. The use of different clustering algorithms or different parameters often produces rather different results on the same data. Biological interpretation of multiple clustering results requires understanding how different clusters relate to each other. It is particularly non-trivial to compare the results of a hierarchical and a flat, e.g. k-means, clustering. RESULTS: We present a new method for comparing and visualizing relationships between different clustering results, either flat versus flat, or flat versus hierarchical. When comparing a flat clustering to a hierarchical clustering, the algorithm cuts different branches in the hierarchical tree at different levels to optimize the correspondence between the clusters. The optimization function is based on graph layout aesthetics or on mutual information. The clusters are displayed using a bipartite graph where the edges are weighted proportionally to the number of common elements in the respective clusters and the weighted number of crossings is minimized. The performance of the algorithm is tested using simulated and real gene expression data. The algorithm is implemented in the online gene expression data analysis tool Expression Profiler. AVAILABILITY: http://www.ebi.ac.uk/expressionprofiler     

Detecting clusters of different geometrical shapes in microarray gene expression data

MOTIVATION: Clustering has been used as a popular technique for finding groups of genes that show similar expression patterns under multiple experimental conditions. Many clustering methods have been proposed for clustering gene-expression data, including the hierarchical clustering, k-means clustering and self-organizing map (SOM). However, the conventional methods are limited to identify different shapes of clusters because they use a fixed distance norm when calculating the distance between genes. The fixed distance norm imposes a fixed geometrical shape on the clusters regardless of the actual data distribution. Thus, different distance norms are required for handling the different shapes of clusters. RESULTS: We present the Gustafson-Kessel (GK) clustering method for microarray gene-expression data. To detect clusters of different shapes in a dataset, we use an adaptive distance norm that is calculated by a fuzzy covariance matrix (F) of each cluster in which the eigenstructure of F is used as an indicator of the shape of the cluster. Moreover, the GK method is less prone to falling into local minima than the k-means and SOM because it makes decisions through the use of membership degrees of a gene to clusters. The algorithmic procedure is accomplished by the alternating optimization technique, which iteratively improves a sequence of sets of clusters until no further improvement is possible. To test the performance of the GK method, we applied the GK method and well-known conventional methods to three recently published yeast datasets, and compared the performance of each method using the Saccharomyces Genome Database annotations. The clustering results of the GK method are more significantly relevant to the biological annotations than those of the other methods, demonstrating its effectiveness and potential for clustering gene-expression data. AVAILABILITY: The software was developed using Java language, and can be executed on the platforms that JVM (Java Virtual Machine) is running. It is available from the authors upon request. SUPPLEMENTARY INFORMATION: Supplementary data are available at http://dragon.kaist.ac.kr/gk.     

Performance assessment of kernel density clustering for gene expression profile data

Kernel density smoothing techniques have been used in classification or supervised learning of gene expression profile (GEP) data, but their applications to clustering or unsupervised learning of those data have not been explored and assessed. Here we report a kernel density clustering method for analysing GEP data and compare its performance with the three most widely-used clustering methods: hierarchical clustering, K-means clustering, and multivariate mixture model-based clustering. Using several methods to measure agreement, between-cluster isolation, and withincluster coherence, such as the Adjusted Rand Index, the Pseudo F test, the r(2) test, and the profile plot, we have assessed the effectiveness of kernel density clustering for recovering clusters, and its robustness against noise on clustering both simulated and real GEP data. Our results show that the kernel density clustering method has excellent performance in recovering clusters from simulated data and in grouping large real expression profile data sets into compact and well-isolated clusters, and that it is the most robust clustering method for analysing noisy expression profile data compared to the other three methods assessed.     

Reducing the Time Requirement of k-Means Algorithm

Traditional k-means and most k-means variants are still computationally expensive for large datasets, such as microarray data, which have large datasets with large dimension size d. In k-means clustering, we are given a set of n data points in d-dimensional space R^d and an integer k. The problem is to determine a set of k points in R^d, called centers, so as to minimize the mean squared distance from each data point to its nearest center. In this work, we develop a novel k-means algorithm, which is simple but more efficient than the traditional k-means and the recent enhanced k-means. Our new algorithm is based on the recently established relationship between principal component analysis and the k-means clustering. We provided the correctness proof for this algorithm. Results obtained from testing the algorithm on three biological data and six non-biological data (three of these data are real, while the other three are simulated) also indicate that our algorithm is empirically faster than other known k-means algorithms. We assessed the quality of our algorithm clusters against the clusters of a known structure using the Hubert-Arabie Adjusted Rand index (ARI_HA). We found that when k is close to d, the quality is good (ARI_HA>0.8) and when k is not close to d, the quality of our new k-means algorithm is excellent (ARI_HA>0.9). In this paper, emphases are on the reduction of the time requirement of the k-means algorithm and its application to microarray data due to the desire to create a tool for clustering and malaria research. However, the new clustering algorithm can be used for other clustering needs as long as an appropriate measure of distance between the centroids and the members is used. This has been demonstrated in this work on six non-biological data.     

Unraveling tobacco BY-2 protein complexes with BN PAGE/LC-MS/MS and clustering methods

To understand physiological processes, insight into protein complexes is very important. Through a combination of blue native gel electrophoresis and LC-MS/MS, we were able to isolate protein complexes and identify their potential subunits from Nicotiana tabacum cv. Bright Yellow-2. For this purpose, a bioanalytical approach was used that works without a priori knowledge of the interacting proteins. Different clustering methods (e.g., k-means and hierarchical clustering) and a biclustering approach were evaluated according to their ability to group proteins by their migration profile and to correlate the proteins to a specific complex. The biclustering approach was identified as a very powerful tool for the exploration of protein complexes of whole cell lysates since it allows for the promiscuous nature of proteins. Furthermore, it searches for associations between proteins that co-occur frequently throughout the BN gel, which increases the confidence of the putative associations between co-migrating proteins. The statistical significance and biological relevance of the profile clusters were verified using functional gene ontology annotation. The proof of concept for identifying protein complexes by our BN PAGE/LC-MS/MS approach is provided through the analysis of known protein complexes. Both well characterized long-lived protein complexes as well as potential temporary sequential multi-enzyme complexes were characterized.     

Comparison of five clustering algorithms to classify phytoplankton from flow cytometry data.

BACKGROUND: Artificial neural networks (ANNs) have been shown to be valuable in the analysis of analytical flow cytometric (AFC) data in aquatic ecology. Automated extraction of clusters is an important first stage in deriving ANN training data from field samples, but AFC data pose a number of challenges for many types of clustering algorithm. The fuzzy k-means algorithm recently has been extended to address nonspherical clusters with the use of scatter matrices. Four variants were proposed, each optimizing a different measure of clustering "goodness." METHODS: With AFC data obtained from marine phytoplankton species in culture, the four fuzzy k-means algorithm variants were compared with each other and with another multivariate clustering algorithm based on critical distances currently used in flow cytometry. RESULTS: One of the algorithm variants (adaptive distances, also known as the Gustafson--Kessel algorithm) was found to be robust and reliable, whereas the others showed various problems. CONCLUSIONS: The adaptive distances algorithm was superior in use to the clustering algorithms against which it was tested, but the problem of automatic determination of the number of clusters remains to be addressed.     

Protein complex prediction via cost-based clustering

MOTIVATION: Understanding principles of cellular organization and function can be enhanced if we detect known and predict still undiscovered protein complexes within the cell's protein-protein interaction (PPI) network. Such predictions may be used as an inexpensive tool to direct biological experiments. The increasing amount of available PPI data necessitates an accurate and scalable approach to protein complex identification. RESULTS: We have developed the Restricted Neighborhood Search Clustering Algorithm (RNSC) to efficiently partition networks into clusters using a cost function. We applied this cost-based clustering algorithm to PPI networks of Saccharomyces cerevisiae, Drosophila melanogaster and Caenorhabditis elegans to identify and predict protein complexes. We have determined functional and graph-theoretic properties of true protein complexes from the MIPS database. Based on these properties, we defined filters to distinguish between identified network clusters and true protein complexes. Conclusions: Our application of the cost-based clustering algorithm provides an accurate and scalable method of detecting and predicting protein complexes within a PPI network.     

A robust approach based on Weibull distribution for clustering gene expression data

Background

Clustering is a widely used technique for analysis of gene expression data. Most clustering methods group genes based on the distances, while few methods group genes according to the similarities of the distributions of the gene expression levels. Furthermore, as the biological annotation resources accumulated, an increasing number of genes have been annotated into functional categories. As a result, evaluating the performance of clustering methods in terms of the functional consistency of the resulting clusters is of great interest.

Results

In this paper, we proposed the WDCM (Weibull Distribution-based Clustering Method), a robust approach for clustering gene expression data, in which the gene expressions of individual genes are considered as the random variables following unique Weibull distributions. Our WDCM is based on the concept that the genes with similar expression profiles have similar distribution parameters, and thus the genes are clustered via the Weibull distribution parameters. We used the WDCM to cluster three cancer gene expression data sets from the lung cancer, B-cell follicular lymphoma and bladder carcinoma and obtained well-clustered results. We compared the performance of WDCM with k-means and Self Organizing Map (SOM) using functional annotation information given by the Gene Ontology (GO). The results showed that the functional annotation ratios of WDCM are higher than those of the other methods. We also utilized the external measure Adjusted Rand Index to validate the performance of the WDCM. The comparative results demonstrate that the WDCM provides the better clustering performance compared to k-means and SOM algorithms. The merit of the proposed WDCM is that it can be applied to cluster incomplete gene expression data without imputing the missing values. Moreover, the robustness of WDCM is also evaluated on the incomplete data sets.

Conclusions

The results demonstrate that our WDCM produces clusters with more consistent functional annotations than the other methods. The WDCM is also verified to be robust and is capable of clustering gene expression data containing a small quantity of missing values.     

Increased expression of the 43-kD protein disrupts acetylcholine receptor clustering in myotubes

The 43-kD protein is a peripheral membrane protein that is in approximately 1:1 stoichiometry with the acetylcholine receptor (AChR) in vertebrate muscle cells and colocalizes with it in the postsynaptic membrane. To investigate the role of the 43-kD protein in AChR clustering, we have isolated C2 muscle cell lines in which some cells overexpress the 43-kD protein. We find that myotubes with increased levels of the 43-kD protein have small AChR clusters and that those with the highest levels of expression have a drastically reduced number of clusters. Our results suggest that the 1:1 stoichiometry of AChR and 43-kD protein found in muscle cells is important for AChR cluster formation.     

Analysis of expression profile using fuzzy adaptive resonance theory

MOTIVATION: It is well understood that the successful clustering of expression profiles give beneficial ideas to understand the functions of uncharacterized genes. In order to realize such a successful clustering, we investigate a clustering method based on adaptive resonance theory (ART) in this report. RESULTS: We apply Fuzzy ART as a clustering method for analyzing the time series expression data during sporulation of Saccharomyces cerevisiae. The clustering result by Fuzzy ART was compared with those by other clustering methods such as hierarchical clustering, k-means algorithm and self-organizing maps (SOMs). In terms of the mathematical validations, Fuzzy ART achieved the most reasonable clustering. We also verified the robustness of Fuzzy ART using noised data. Furthermore, we defined the correctness ratio of clustering, which is based on genes whose temporal expressions are characterized biologically. Using this definition, it was proved that the clustering ability of Fuzzy ART was superior to other clustering methods such as hierarchical clustering, k-means algorithm and SOMs. Finally, we validate the clustering results by Fuzzy ART in terms of biological functions and evidence. AVAILABILITY: The software is available at http//www.nubio.nagoya-u.ac.jp/proc/index.html     

A dynamically growing self-organizing tree (DGSOT) for hierarchical clustering gene expression profiles

MOTIVATION: The increasing use of microarray technologies is generating large amounts of data that must be processed in order to extract useful and rational fundamental patterns of gene expression. Hierarchical clustering technology is one method used to analyze gene expression data, but traditional hierarchical clustering algorithms suffer from several drawbacks (e.g. fixed topology structure; mis-clustered data which cannot be reevaluated). In this paper, we introduce a new hierarchical clustering algorithm that overcomes some of these drawbacks. RESULT: We propose a new tree-structure self-organizing neural network, called dynamically growing self-organizing tree (DGSOT) algorithm for hierarchical clustering. The DGSOT constructs a hierarchy from top to bottom by division. At each hierarchical level, the DGSOT optimizes the number of clusters, from which the proper hierarchical structure of the underlying dataset can be found. In addition, we propose a new cluster validation criterion based on the geometric property of the Voronoi partition of the dataset in order to find the proper number of clusters at each hierarchical level. This criterion uses the Minimum Spanning Tree (MST) concept of graph theory and is computationally inexpensive for large datasets. A K-level up distribution (KLD) mechanism, which increases the scope of data distribution in the hierarchy construction, was used to improve the clustering accuracy. The KLD mechanism allows the data misclustered in the early stages to be reevaluated at a later stage and increases the accuracy of the final clustering result. The clustering result of the DGSOT is easily displayed as a dendrogram for visualization. Based on a yeast cell cycle microarray expression dataset, we found that our algorithm extracts gene expression patterns at different levels. Furthermore, the biological functionality enrichment in the clusters is considerably high and the hierarchical structure of the clusters is more reasonable. AVAILABILITY: DGSOT is available upon request from the authors.     

Annotating gene function by combining expression data with a modular gene network

MOTIVATION: A promising and reliable approach to annotate gene function is clustering genes not only by using gene expression data but also literature information, especially gene networks. RESULTS: We present a systematic method for gene clustering by combining these totally different two types of data, particularly focusing on network modularity, a global feature of gene networks. Our method is based on learning a probabilistic model, which we call a hidden modular random field in which the relation between hidden variables directly represents a given gene network. Our learning algorithm which minimizes an energy function considering the network modularity is practically time-efficient, regardless of using the global network property. We evaluated our method by using a metabolic network and microarray expression data, changing with microarray datasets, parameters of our model and gold standard clusters. Experimental results showed that our method outperformed other four competing methods, including k-means and existing graph partitioning methods, being statistically significant in all cases. Further detailed analysis showed that our method could group a set of genes into a cluster which corresponds to the folate metabolic pathway while other methods could not. From these results, we can say that our method is highly effective for gene clustering and annotating gene function.     

Using indirect protein-protein interactions for protein complex prediction

Protein complexes are fundamental for understanding principles of cellular organizations. As the sizes of protein-protein interaction (PPI) networks are increasing, accurate and fast protein complex prediction from these PPI networks can serve as a guide for biological experiments to discover novel protein complexes. However, it is not easy to predict protein complexes from PPI networks, especially in situations where the PPI network is noisy and still incomplete. Here, we study the use of indirect interactions between level-2 neighbors (level-2 interactions) for protein complex prediction. We know from previous work that proteins which do not interact but share interaction partners (level-2 neighbors) often share biological functions. We have proposed a method in which all direct and indirect interactions are first weighted using topological weight (FS-Weight), which estimates the strength of functional association. Interactions with low weight are removed from the network, while level-2 interactions with high weight are introduced into the interaction network. Existing clustering algorithms can then be applied to this modified network. We have also proposed a novel algorithm that searches for cliques in the modified network, and merge cliques to form clusters using a "partial clique merging" method. Experiments show that (1) the use of indirect interactions and topological weight to augment protein-protein interactions can be used to improve the precision of clusters predicted by various existing clustering algorithms; and (2) our complex-finding algorithm performs very well on interaction networks modified in this way. Since no other information except the original PPI network is used, our approach would be very useful for protein complex prediction, especially for prediction of novel protein complexes.     

Using repeated measurements to validate hierarchical gene clusters

MOTIVATION: Hierarchical clustering is a common approach to study protein and gene expression data. This unsupervised technique is used to find clusters of genes or proteins which are expressed in a coordinated manner across a set of conditions. Because of both the biological and technical variability, experimental repetitions are generally performed. In this work, we propose an approach to evaluate the stability of clusters derived from hierarchical clustering by taking repeated measurements into account. RESULTS: The method is based on the bootstrap technique that is used to obtain pseudo-hierarchies of genes from resampled datasets. Based on a fast dynamic programming algorithm, we compare the original hierarchy to the pseudo-hierarchies and assess the stability of the original gene clusters. Then a shuffling procedure can be used to assess the significance of the cluster stabilities. Our approach is illustrated on simulated data and on two microarray datasets. Compared to the standard hierarchical clustering methodology, it allows to point out the dubious and stable clusters, and thus avoids misleading interpretations. AVAILABILITY: The programs were developed in C and R languages.     

Effective algorithm for determining the number of clusters and its application in image segmentation

The k-means algorithm is a popular clustering method for image segmentation. However, the main disadvantage of this algorithm is its dependence on the number of initial clusters. In this paper, we present an optimal criterion which can select the best segmentation result with less number of clusters. The optimal criterion overcomes the shortcoming of initialization based on the intra-class and inter-class difference. Eight digital images were employed to verify the segmentation results of the optimal criterion. Simultaneously, we have improved the traditional k-means algorithm to find the initial clustering centers efficiently. Experimental results show that the segmented images selected by the optimal criterion have sufficient stability and robustness. In addition, we verify the consistency of results by two kinds of objective assessment measures. The proposed optimal criterion can successfully display the best segmentation results precisely and efficiently so as to instead of artificial selection.