Protein Similarity Networks Research Articles

Exploring Biotic Interactions Within Protist Cell Populations Using Network Methods

The study of diseased human cells and of cells isolated from the natural environment will likely be revolutionized by single cell genomics (SCG). Here, we used protein similarity networks to explore within- and between-cell DNA differences from SCG data derived from six individual rhizarian cells related to Paulinella ovalis and proteins from the complete genome of another rhizarian, Bigelowiella natans. We identified shared and distinct DNA components within our SCG data and between P. ovalis and B. natans. We show that network properties such as assortativity and degree effectively discriminate genome features between SCG assemblies and that SCG data follow the power law with a small number of protein families dominating networks.

Diversity of the metal-transporting P1B-type ATPases.

The P1B-ATPases are integral membrane proteins that couple ATP hydrolysis to metal cation transport. Widely distributed across all domains of life, these enzymes have been previously shown to transport copper, zinc, cobalt, and other thiophilic heavy metals. Recent data suggest that these enzymes may also be involved in nickel and/or iron transport. Here we have exploited large amounts of genomic data to examine and classify the various P1B-ATPase subfamilies. Specifically, we have combined new methods of data partitioning and network visualization known as Transitivity Clustering and Protein Similarity Networks with existing biochemical data to examine properties such as length, speciation, and metal-binding motifs of the P1B-ATPase subfamily sequences. These data reveal interesting relationships among the enzyme sequences of previously established subfamilies, indicate the presence of two new subfamilies, and suggest the existence of new regulatory elements in certain subfamilies. Taken together, these findings underscore the importance of P1B-ATPases in homeostasis of nearly every biologically relevant transition metal and provide an updated framework for future studies.

PANADA: Protein Association Network Annotation, Determination and Analysis

Increasingly large numbers of proteins require methods for functional annotation. This is typically based on pairwise inference from the homology of either protein sequence or structure. Recently, similarity networks have been presented to leverage both the ability to visualize relationships between proteins and assess the transferability of functional inference. Here we present PANADA, a novel toolkit for the visualization and analysis of protein similarity networks in Cytoscape. Networks can be constructed based on pairwise sequence or structural alignments either on a set of proteins or, alternatively, by database search from a single sequence. The Panada web server, executable for download and examples and extensive help files are available at URL: http://protein.bio.unipd.it/panada/.

SIMAP—the database of all-against-all protein sequence similarities and annotations with new interfaces and increased coverage

The Similarity Matrix of Proteins (SIMAP, http://mips.gsf.de/simap/) database has been designed to massively accelerate computationally expensive protein sequence analysis tasks in bioinformatics. It provides pre-calculated sequence similarities interconnecting the entire known protein sequence universe, complemented by pre-calculated protein features and domains, similarity clusters and functional annotations. SIMAP covers all major public protein databases as well as many consistently re-annotated metagenomes from different repositories. As of September 2013, SIMAP contains >163 million proteins corresponding to ∼70 million non-redundant sequences. SIMAP uses the sensitive FASTA search heuristics, the Smith–Waterman alignment algorithm, the InterPro database of protein domain models and the BLAST2GO functional annotation algorithm. SIMAP assists biologists by facilitating the interactive exploration of the protein sequence universe. Web-Service and DAS interfaces allow connecting SIMAP with any other bioinformatic tool and resource. All-against-all protein sequence similarity matrices of project-specific protein collections are generated on request. Recent improvements allow SIMAP to cover the rapidly growing sequenced protein sequence universe. New Web-Service interfaces enhance the connectivity of SIMAP. Novel tools for interactive extraction of protein similarity networks have been added. Open access to SIMAP is provided through the web portal; the portal also contains instructions and links for software access and flat file downloads.

Protein Similarity Networks Reveal Relationships among Sequence, Structure, and Function within the Cupin Superfamily

The cupin superfamily is extremely diverse and includes catalytically inactive seed storage proteins, sugar-binding metal-independent epimerases, and metal-dependent enzymes possessing dioxygenase, decarboxylase, and other activities. Although numerous proteins of this superfamily have been structurally characterized, the functions of many of them have not been experimentally determined. We report the first use of protein similarity networks (PSNs) to visualize trends of sequence and structure in order to make functional inferences in this remarkably diverse superfamily. PSNs provide a way to visualize relatedness of structure and sequence among a given set of proteins. Structure- and sequence-based clustering of cupin members reflects functional clustering. Networks based only on cupin domains and networks based on the whole proteins provide complementary information. Domain-clustering supports phylogenetic conclusions that the N- and C-terminal domains of bicupin proteins evolved independently. Interestingly, although many functionally similar enzymatic cupin members bind the same active site metal ion, the structure and sequence clustering does not correlate with the identity of the bound metal. It is anticipated that the application of PSNs to this superfamily will inform experimental work and influence the functional annotation of databases.

A Global Comparison of the Human and T. brucei Degradomes Gives Insights about Possible Parasite Drug Targets

We performed a genome-level computational study of sequence and structure similarity, the latter using crystal structures and models, of the proteases of Homo sapiens and the human parasite Trypanosoma brucei. Using sequence and structure similarity networks to summarize the results, we constructed global views that show visually the relative abundance and variety of proteases in the degradome landscapes of these two species, and provide insights into evolutionary relationships between proteases. The results also indicate how broadly these sequence sets are covered by three-dimensional structures. These views facilitate cross-species comparisons and offer clues for drug design from knowledge about the sequences and structures of potential drug targets and their homologs. Two protease groups (“M32” and “C51”) that are very different in sequence from human proteases are examined in structural detail, illustrating the application of this global approach in mining new pathogen genomes for potential drug targets. Based on our analyses, a human ACE2 inhibitor was selected for experimental testing on one of these parasite proteases, TbM32, and was shown to inhibit it. These sequence and structure data, along with interactive versions of the protein similarity networks generated in this study, are available at http://babbittlab.ucsf.edu/resources.html.

Pythoscape: a framework for generation of large protein similarity networks

Pythoscape is a framework implemented in Python for processing large protein similarity networks for visualization in other software packages. Protein similarity networks are graphical representations of sequence, structural and other similarities among proteins for which pairwise all-by-all similarity connections have been calculated. Mapping of biological and other information to network nodes or edges enables hypothesis creation about sequence–structure–function relationships across sets of related proteins. Pythoscape provides several options to calculate pairwise similarities for input sequences or structures, applies filters to network edges and defines sets of similar nodes and their associated data as single nodes (termed representative nodes) for compression of network information and output data or formatted files for visualization.Contact: babbitt@cgl.ucsf.eduSupplementary information: Supplementary data are available at Bioinformatics online.

Detecting Network Communities: An Application to Phylogenetic Analysis

This paper proposes a new method to identify communities in generally weighted complex networks and apply it to phylogenetic analysis. In this case, weights correspond to the similarity indexes among protein sequences, which can be used for network construction so that the network structure can be analyzed to recover phylogenetically useful information from its properties. The analyses discussed here are mainly based on the modular character of protein similarity networks, explored through the Newman-Girvan algorithm, with the help of the neighborhood matrix . The most relevant networks are found when the network topology changes abruptly revealing distinct modules related to the sets of organisms to which the proteins belong. Sound biological information can be retrieved by the computational routines used in the network approach, without using biological assumptions other than those incorporated by BLAST. Usually, all the main bacterial phyla and, in some cases, also some bacterial classes corresponded totally (100%) or to a great extent (>70%) to the modules. We checked for internal consistency in the obtained results, and we scored close to 84% of matches for community pertinence when comparisons between the results were performed. To illustrate how to use the network-based method, we employed data for enzymes involved in the chitin metabolic pathway that are present in more than 100 organisms from an original data set containing 1,695 organisms, downloaded from GenBank on May 19, 2007. A preliminary comparison between the outcomes of the network-based method and the results of methods based on Bayesian, distance, likelihood, and parsimony criteria suggests that the former is as reliable as these commonly used methods. We conclude that the network-based method can be used as a powerful tool for retrieving modularity information from weighted networks, which is useful for phylogenetic analysis.

Detecting remote evolutionary relationships among proteins by large-scale semantic embedding.

Virtually every molecular biologist has searched a protein or DNA sequence database to find sequences that are evolutionarily related to a given query. Pairwise sequence comparison methods—i.e., measures of similarity between query and target sequences—provide the engine for sequence database search and have been the subject of 30 years of computational research. For the difficult problem of detecting remote evolutionary relationships between protein sequences, the most successful pairwise comparison methods involve building local models (e.g., profile hidden Markov models) of protein sequences. However, recent work in massive data domains like web search and natural language processing demonstrate the advantage of exploiting the global structure of the data space. Motivated by this work, we present a large-scale algorithm called ProtEmbed, which learns an embedding of protein sequences into a low-dimensional “semantic space.” Evolutionarily related proteins are embedded in close proximity, and additional pieces of evidence, such as 3D structural similarity or class labels, can be incorporated into the learning process. We find that ProtEmbed achieves superior accuracy to widely used pairwise sequence methods like PSI-BLAST and HHSearch for remote homology detection; it also outperforms our previous RankProp algorithm, which incorporates global structure in the form of a protein similarity network. Finally, the ProtEmbed embedding space can be visualized, both at the global level and local to a given query, yielding intuition about the structure of protein sequence space.

Improving the quality of protein similarity network clustering algorithms using the network edge weight distribution

Clustering protein sequence data into functionally specific families is a difficult but important problem in biological research. One useful approach for tackling this problem involves representing the sequence dataset as a protein similarity network, and afterwards clustering the network using advanced graph analysis techniques. Although a multitude of such network clustering algorithms have been developed over the past few years, comparing algorithms is often difficult because performance is affected by the specifics of network construction. We investigate an important aspect of network construction used in analyzing protein superfamilies and present a heuristic approach for improving the performance of several algorithms. We analyzed how the performance of network clustering algorithms relates to thresholding the network prior to clustering. Our results, over four different datasets, show how for each input dataset there exists an optimal threshold range over which an algorithm generates its most accurate clustering output. Our results further show how the optimal threshold range correlates with the shape of the edge weight distribution for the input similarity network. We used this correlation to develop an automated threshold selection heuristic in order to most optimally filter a similarity network prior to clustering. This heuristic allows researchers to process their protein datasets with runtime efficient network clustering algorithms without sacrificing the clustering accuracy of the final results. Python code for implementing the automated threshold selection heuristic, together with the datasets used in our analysis, are available at http://www.rbvi.ucsf.edu/Research/cytoscape/threshold_scripts.zip.

A similarity network approach for the analysis and comparison of protein sequence/structure sets

A set of proteins is a complex system whose elements are interrelated on the concept of sequence- and structure-based similarity. Here, we applied a similarity network-based methodology for the representation and analysis of protein sequences and structures sets using a non-redundant set of 311 proteins and three different information criteria based on sequence-derived features, sequence local alignment and structural alignment. A wide set of measurements, like network degree, clustering coefficient, characteristic path length and vertex centrality were utilized to characterize the networks’ topology. Protein similarity networks were found medium or highly interconnected and the existence of both clusters and random edges classified their fully connected versions as Small World Networks (SWNs). The SWN architecture was able to host the continuous similarity transition among proteins and model the protein information flow during evolution. Recently reported ancestral elements, like the α/β class and certain folds, were remarkably found to act as hubs in the networks. Additionally, the moderate information value of sequence-derived features when used for fold and class assignment was shown on a network basis. The methodology described here can be applied for the analysis of other complex systems which consist of interrelated elements and a certain information flow.

An Atlas of the Thioredoxin Fold Class Reveals the Complexity of Function-Enabling Adaptations

The group of proteins that contain a thioredoxin (Trx) fold is huge and diverse. Assessment of the variation in catalytic machinery of Trx fold proteins is essential in providing a foundation for understanding their functional diversity and predicting the function of the many uncharacterized members of the class. The proteins of the Trx fold class retain common features—including variations on a dithiol CxxC active site motif—that lead to delivery of function. We use protein similarity networks to guide an analysis of how structural and sequence motifs track with catalytic function and taxonomic categories for 4,082 representative sequences spanning the known superfamilies of the Trx fold. Domain structure in the fold class is varied and modular, with 2.8% of sequences containing more than one Trx fold domain. Most member proteins are bacterial. The fold class exhibits many modifications to the CxxC active site motif—only 56.8% of proteins have both cysteines, and no functional groupings have absolute conservation of the expected catalytic motif. Only a small fraction of Trx fold sequences have been functionally characterized. This work provides a global view of the complex distribution of domains and catalytic machinery throughout the fold class, showing that each superfamily contains remnants of the CxxC active site. The unifying context provided by this work can guide the comparison of members of different Trx fold superfamilies to gain insight about their structure-function relationships, illustrated here with the thioredoxins and peroxiredoxins.

Rankprop: a web server for protein remote homology detection

Summary: We present a large-scale implementation of the Rankprop protein homology ranking algorithm in the form of an openly accessible web server. We use the NRDB40 PSI-BLAST all-versus-all protein similarity network of 1.1 million proteins to construct the graph for the Rankprop algorithm, whereas previously, results were only reported for a database of 108 000 proteins. We also describe two algorithmic improvements to the original algorithm, including propagation from multiple homologs of the query and better normalization of ranking scores, that lead to higher accuracy and to scores with a probabilistic interpretation.Availability: The Rankprop web server and source code are available at http://rankprop.gs.washington.eduContact: iain@nec-labs.com; noble@gs.washington.edu

Component Evolution in Random Intersection Graphs

We study the evolution of the order of the largest component in the random intersection graph model which reflects some clustering properties of real–world networks. We show that for appropriate choice of the parameters random intersection graphs differ from $G_{n,p}$ in that neither the so-called giant component, appearing when the expected vertex degree gets larger than one, has linear order nor is the second largest of logarithmic order. We also describe a test of our result on a protein similarity network.

Protein ranking by semi-supervised network propagation.

BackgroundBiologists regularly search DNA or protein databases for sequences that share an evolutionary or functional relationship with a given query sequence. Traditional search methods, such as BLAST and PSI-BLAST, focus on detecting statistically significant pairwise sequence alignments and often miss more subtle sequence similarity. Recent work in the machine learning community has shown that exploiting the global structure of the network defined by these pairwise similarities can help detect more remote relationships than a purely local measure.MethodsWe review RankProp, a ranking algorithm that exploits the global network structure of similarity relationships among proteins in a database by performing a diffusion operation on a protein similarity network with weighted edges. The original RankProp algorithm is unsupervised. Here, we describe a semi-supervised version of the algorithm that uses labeled examples. Three possible ways of incorporating label information are considered: (i) as a validation set for model selection, (ii) to learn a new network, by choosing which transfer function to use for a given query, and (iii) to estimate edge weights, which measure the probability of inferring structural similarity.ResultsBenchmarked on a human-curated database of protein structures, the original RankProp algorithm provides significant improvement over local network search algorithms such as PSI-BLAST. Furthermore, we show here that labeled data can be used to learn a network without any need for estimating parameters of the transfer function, and that diffusion on this learned network produces better results than the original RankProp algorithm with a fixed network.ConclusionIn order to gain maximal information from a network, labeled and unlabeled data should be used to extract both local and global structure.

Identifying remote protein homologs by network propagation

Perhaps the most widely used applications of bioinformatics are tools such as psi-blast for searching sequence databases. We describe a recently developed protein database search algorithm called rankprop. rankprop relies upon a precomputed network of pairwise protein similarities. The algorithm performs a diffusion operation from a specified query protein across the protein similarity network. The resulting activation scores, assigned to each database protein, encode information about the global structure of the protein similarity network. This type of algorithm has a rich history in associationist psychology, artificial intelligence and web search. We describe the rankprop algorithm and its relatives, and we provide evidence that the algorithm successfully improves upon the rankings produced by psi-blast.

Protein ranking: from local to global structure in the protein similarity network.

Biologists regularly search databases of DNA or protein sequences for evolutionary or functional relationships to a given query sequence. We describe a ranking algorithm that exploits the entire network structure of similarity relationships among proteins in a sequence database by performing a diffusion operation on a precomputed, weighted network. The resulting ranking algorithm, evaluated by using a human-curated database of protein structures, is efficient and provides significantly better rankings than a local network search algorithm such as psi-blast.

PsbD sequences of Bumilleriopsis filiformis (Heterokontophyta, Xanthophyceae) and Porphyridium purpureum (Rhodophyta, Bangiophycidae): evidence for polyphyletic origins of plastids.

The nucleotide sequences of the plastidal psbD genes of Bumilleriopsis filiformis and Porphyridium purpureum (encoding the D2 protein of photosystem II) are reported in this paper. The Bumilleriopsis sequence clusters together with Porphyridium when a most parsimonious protein tree of D2 sequences is constructed. A composite D1/D2 protein-similarity network reveals that neither the three red algal sequences nor the two heterokontophyte sequences (Bumilleriopsis, xanthophytes and Ectocarpus, phaeophytes) group together. Therefore, the Heterokontophyta and Rhodophyta may be heterogeneous groups. Instead, it emerges that the D1/D2 proteins of Porphyridium and Bumilleriopsis clearly form a tight cluster. D1 and D2 proteins apparently do not provide a reliable molecular clock. These results fit into hypotheses proposing a polyphyletic origin for complex plastids, even among the supposedly "natural" group of heterokontophytes.