Protein Sequence Families Research Articles

A Lazy Data Mining Approach for Protein Classification

In this work, we propose a new computational technique to solve the protein classification problem. The goal is to predict the functional family of novel protein sequences based on their motif composition. In order to improve the results obtained with other known approaches, we propose a new data mining technique for protein classification based on Bayes' theorem, called highest subset probability (HiSP). To evaluate our proposal, datasets extracted from Prosite, a curated protein family database, are used as experimental datasets. The computational results have shown that the proposed method outperforms other known methods for all tested datasets and looks very promising for problems with characteristics similar to the problem addressed here. In addition, our experiments suggest that HiSP performs well on highly imbalanced datasets.

A Structural Viral Mimic of Prosurvival Bcl-2: A Pivotal Role for Sequestering Proapoptotic Bax and Bak

Many viruses express antiapoptotic proteins to counter host defense mechanisms that would otherwise trigger the rapid clearance of infected cells. For example, adenoviruses and some gamma-herpesviruses express homologs of prosurvival Bcl-2 to subvert the host's apoptotic machinery. Myxoma virus, a double-stranded DNA virus of the pox family, harbors antiapoptotic M11L, its virulence factor. Analysis of its three-dimensional structure reveals that despite lacking any primary sequence similarity to Bcl-2, it adopts a virtually identical protein fold. This allows it to associate with BH3 domains, especially those of Bax and Bak. We found that M11L acts primarily by sequestering Bax and Bak, thereby blocking the killing action of these essential cell-death mediators. These findings expand the family of protein sequences that act like Bcl-2 to block apoptosis and support the conclusion that the prosurvival action of these proteins critically depends on their ability to bind and antagonize Bax and/or Bak.

The structure ofPlasmodium vivaxphosphatidylethanolamine-binding protein suggests a functional motif containing a left-handed helix

The structure of a putative Raf kinase inhibitor protein (RKIP) homolog from the eukaryotic parasite Plasmodium vivax has been studied to a resolution of 1.3 A using multiple-wavelength anomalous diffraction at the Se K edge. This protozoan protein is topologically similar to previously studied members of the phosphatidylethanolamine-binding protein (PEBP) sequence family, but exhibits a distinctive left-handed alpha-helical region at one side of the canonical phospholipid-binding site. Re-examination of previously determined PEBP structures suggests that the P. vivax protein and yeast carboxypeptidase Y inhibitor may represent a structurally distinct subfamily of the diverse PEBP-sequence family.

ProFaM: An Efficient Algorithm for Protein Sequence Family Mining

ProFaM: An Efficient Algorithm for Protein Sequence Family Mining

Improvement in accuracy of multiple sequence alignment using novel group-to-group sequence alignment algorithm with piecewise linear gap cost

BackgroundMultiple sequence alignment (MSA) is a useful tool in bioinformatics. Although many MSA algorithms have been developed, there is still room for improvement in accuracy and speed. In the alignment of a family of protein sequences, global MSA algorithms perform better than local ones in many cases, while local ones perform better than global ones when some sequences have long insertions or deletions (indels) relative to others. Many recent leading MSA algorithms have incorporated pairwise alignment information obtained from a mixture of sources into their scoring system to improve accuracy of alignment containing long indels.ResultsWe propose a novel group-to-group sequence alignment algorithm that uses a piecewise linear gap cost. We developed a program called PRIME, which employs our proposed algorithm to optimize the well-defined sum-of-pairs score. PRIME stands for Profile-based Randomized Iteration MEthod. We evaluated PRIME and some recent MSA programs using BAliBASE version 3.0 and PREFAB version 4.0 benchmarks. The results of benchmark tests showed that PRIME can construct accurate alignments comparable to the most accurate programs currently available, including L-INS-i of MAFFT, ProbCons, and T-Coffee.ConclusionPRIME enables users to construct accurate alignments without having to employ pairwise alignment information. PRIME is available at .

Evolution of protein structural classes and protein sequence families

In protein structure space, protein structures cluster into four elongated regions when mapped based solely on similarity among the 3D structures. These four regions correspond to the four major classes of present-day proteins defined by the contents of secondary structure types and their topological arrangement. Evolution of and restriction to these four classes suggest that, in most cases, the evolution of genes may have been constrained or selected to those genetic changes that results in structurally stable proteins occupying one of the four "allowed" regions of the protein structure space, "structural selection," an important component of natural selection in gene evolution. Our studies on tracing the "common structural ancestor" for each protein sequence family of known structure suggest that: (i) recently emerged proteins belong mostly to three classes; (ii) the proteins that emerged earlier evolved to gain a new class; and (iii) the proteins that emerged earliest evolved to become the present-day proteins in the four major classes, with the fourth-class proteins becoming the most dominant population. Furthermore, our studies also show that not all present-day proteins evolved from one single set of proteins in the last common ancestral organism, but new common ancestral proteins were "born" at different evolutionary times, not traceable to one or two ancestral proteins: "the multiple birth model" for the evolution of protein sequence families.

REMUS: a tool for identification of unique peptide segments as epitopes

We provide a ‘REMUS’ (reinforced merging techniques for unique peptide segments) web server for identification of the locations and compositions of unique peptide segments from a set of protein family sequences. Different levels of uniqueness are determined according to substitutional relationship in the amino acids, frequency of appearance and biological properties such as priority for serving as candidates for epitopes where antibodies recognize. REMUS also provides interactive visualization of 3D structures for allocation and comparison of the identified unique peptide segments. Accuracy of the algorithm was found to be 70% in terms of mapping a unique peptide segment as an epitope. The REMUS web server is available at and the PC version software can be freely downloaded either at or . User guide and working examples for PC version are available at , and details of the proposed algorithm can be referred to the documents as described previously [H. T. Chang, T. W. Pai, T. C. Fan, B. H. Su, P. C. Wu, C. Y. Tang, C. T. Chang, S. H. Liu and M. D. T. Chang (2006) BMC Bioinformatics, 7, 38 and T. W. Pai, B. H. Su, P. C. Wu, M. D. T. Chang, H. T. Chang, T. C. Fan and S. H. Liu (2006) J. Bioinform. Comput. Biol., 4, 75–92].

Identification of latent periodicity in amino acid sequences of protein families

For detection of the latent periodicity of the protein families responsible for various biological functions, methods of information decomposition, cyclic profile alignment, and the method of noise decomposition have been used. The latent periodicity, being specific to a particular family, is recognized in 94 of 110 analyzed protein families. Family specific periodicity was found for more than 70% of amino acid sequences in each of these families. Based on such sequences the characteristic profile of the latent periodicity has been deduced for each family. Possible relationship between the recognized latent periodicity, evolution of proteins, and their structural organization is discussed.

Specificity prediction of adenylation domains in nonribosomal peptide synthetases (NRPS) using transductive support vector machines (TSVMs)

We present a new support vector machine (SVM)-based approach to predict the substrate specificity of subtypes of a given protein sequence family. We demonstrate the usefulness of this method on the example of aryl acid-activating and amino acid-activating adenylation domains (A domains) of nonribosomal peptide synthetases (NRPS). The residues of gramicidin synthetase A that are 8 Å around the substrate amino acid and corresponding positions of other adenylation domain sequences with 397 known and unknown specificities were extracted and used to encode this physico-chemical fingerprint into normalized real-valued feature vectors based on the physico-chemical properties of the amino acids. The SVM software package SVMlight was used for training and classification, with transductive SVMs to take advantage of the information inherent in unlabeled data. Specificities for very similar substrates that frequently show cross-specificities were pooled to the so-called composite specificities and predictive models were built for them. The reliability of the models was confirmed in cross-validations and in comparison with a currently used sequence-comparison-based method. When comparing the predictions for 1230 NRPS A domains that are currently detectable in UniProt, the new method was able to give a specificity prediction in an additional 18% of the cases compared with the old method. For 70% of the sequences both methods agreed, for <6% they did not, mainly on low-confidence predictions by the existing method. None of the predictive methods could infer any specificity for 2.4% of the sequences, suggesting completely new types of specificity.

SIMPROT: Using an empirically determined indel distribution in simulations of protein evolution

BackgroundGeneral protein evolution models help determine the baseline expectations for the evolution of sequences, and they have been extensively useful in sequence analysis and for the computer simulation of artificial sequence data sets.ResultsWe have developed a new method of simulating protein sequence evolution, including insertion and deletion (indel) events in addition to amino-acid substitutions. The simulation generates both the simulated sequence family and a true sequence alignment that captures the evolutionary relationships between amino acids from different sequences. Our statistical model for indel evolution is based on the empirical indel distribution determined by Qian and Goldstein. We have parameterized this distribution so that it applies to sequences diverged by varying evolutionary times and generalized it to provide flexibility in simulation conditions. Our method uses a Monte-Carlo simulation strategy, and has been implemented in a C++ program named Simprot.ConclusionSimprot will be useful for testing methods of analysis of protein sequence families particularly alignment methods, phylogenetic tree building, detection of recombination and horizontal gene transfer, and homology detection, where knowing the true course of sequence evolution is essential.

Protein Family Clustering for Structural Genomics

A major goal of structural genomics is the provision of a structural template for a large fraction of protein domains. The magnitude of this task depends on the number and nature of protein sequence families. With a large number of bacterial genomes now fully sequenced, it is possible to obtain improved estimates of the number and diversity of families in that kingdom. We have used an automated clustering procedure to group all sequences in a set of genomes into protein families. Bench-marking shows the clustering method is sensitive at detecting remote family members, and has a low level of false positives. This comprehensive protein family set has been used to address the following questions. (1) What is the structure coverage for currently known families? (2) How will the number of known apparent families grow as more genomes are sequenced? (3) What is a practical strategy for maximizing structure coverage in future? Our study indicates that approximately 20% of known families with three or more members currently have a representative structure. The study indicates also that the number of apparent protein families will be considerably larger than previously thought: We estimate that, by the criteria of this work, there will be about 250,000 protein families when 1000 microbial genomes have been sequenced. However, the vast majority of these families will be small, and it will be possible to obtain structural templates for 70-80% of protein domains with an achievable number of representative structures, by systematically sampling the larger families.

MAVL/StickWRLD for protein: visualizing protein sequence families to detect non-consensus features

A fundamental problem with applying Consensus, Weight-Matrix or hidden Markov models as search tools for biosequences is that there is no way to know, from the model, if the modeled sequences display any dependencies between positional identities. In some instances, these dependencies are crucial in correctly accepting or rejecting other sequences as members of the family. MAVL (multiple alignment variation linker) and StickWRLD provide a web-based method to visually survey the model-training sequences to discover and characterize possible dependencies. Initially introduced for nucleic acid sequences, with MAVL/StickWRLD, it is easy to distinguish typical DNA or RNA structural dependencies in input families, identify mixed populations of distinct subfamilies, or discover novel dependencies that result from binding interactions or other selective pressures [W. Ray (2004) Nucleic Acids Res., 32, W59–W63]. Since the announcement of MAVL/StickWRLD for nucleic acids, one of the most requested new features has been the extension of this visualization method to support protein alignments. We are pleased to report that this extension has been successful, that the basic visualization has been augmented in several ways to enhance protein viewing, and that the results with protein alignments are even more dramatic than with NA alignments. MAVL/StickWRLD can be accessed at .

Location proteomics: a systems approach to subcellular location

Systems Biology requires comprehensive systematic data on all aspects and levels of biological organization and function. In addition to information on the sequence, structure, activities and binding interactions of all biological macromolecules, the creation of accurate predictive models of cell behaviour will require detailed information on the distribution of those molecules within cells and the ways in which those distributions change over the cell cycle and in response to mutations or external stimuli. Current information on subcellular location in protein databases is limited to unstructured text descriptions or sets of terms assigned by human curators. These entries do not permit basic operations that are common to other biological databases, such as measurement of the degree of similarity between the distributions of two proteins, and they are not able to fully capture the complexity of protein patterns that can be observed. The field of location proteomics seeks to provide automated, objective high-resolution descriptions of protein location patterns within cells. Methods have been developed to group proteins into statistically indistinguishable location patterns using automated analysis of fluorescence microscope images. The resulting clusters, or location families, are analogous to clusters found for other domains, such as protein sequence families. Preliminary work suggests the feasibility of expressing each unique pattern as a generative model that can be incorporated into comprehensive models of cell behaviour.

Conservative extraction of over-represented extensible motifs

The discovery of motifs in biosequences is frequently torn between the rigidity of the model on the one hand and the abundance of candidates on the other. In particular, the variety of motifs described by strings that include 'don't care' (dot) patterns escalates exponentially with the length of the motif, and this gets only worse if a dot is allowed to stretch up to some prescribed maximum length. This circumstance tends to generate daunting computational burdens, and often gives rise to tables that are impossible to visualize and digest. This is unfortunate, as it seems to preclude precisely those massive analyses that have become conceivable with the increasing availability of massive genomic and protein data. Although a part of the problem is endemic, another part of it seems rooted in the various characterizations offered for the notion of a motif, that are typically based either on syntax or on statistics alone. It seems worthwhile to consider alternatives that result from a prudent combination of these two aspects in the model. We introduce and study a notion of extensible motif in a sequence which tightly combines the structure of the motif pattern, as described by its syntactic specification, with the statistical measure of its occurrence count. We show that a combination of appropriate saturation conditions (expressed in terms of minimum number of dots compatible with a given list of occurrences) and the monotonicity of probabilistic scores over regions of constant frequency afford us significant parsimony in the generation and testing of candidate over-represented motifs. The merits of the method are documented by the results obtained in implementation, which specifically targeted protein sequence families. In all cases tested, the motif reported in PROSITE as the most important in terms of functional/structural relevance emerges among the top 30 extensible motifs returned by our algorithm, often right at the top. Of equal importance seems the fact that the sets of all surprising motifs returned in each experiment are extracted faster and come in much more manageable sizes than would be obtained in the absence of saturation constrains. This software will be available for use with the suite of tools at www.research.ibm.com/bioinformatics.

Progress of Structural Genomics Initiatives: An Analysis of Solved Target Structures

The explosion in gene sequence data and technological breakthroughs in protein structure determination inspired the launch of structural genomics (SG) initiatives. An often stated goal of structural genomics is the high-throughput structural characterisation of all protein sequence families, with the long-term hope of significantly impacting on the life sciences, biotechnology and drug discovery. Here, we present a comprehensive analysis of solved SG targets to assess progress of these initiatives. Eleven consortia have contributed 316 non-redundant entries and 323 protein chains to the Protein Data Bank (PDB), and 459 and 393 domains to the CATH and SCOP structure classifications, respectively. The quality and size of these proteins are comparable to those solved in traditional structural biology and, despite huge scope for duplicated efforts, only 14% of targets have a close homologue (>/=30% sequence identity) solved by another consortium. Analysis of CATH and SCOP revealed the significant contribution that structural genomics is making to the coverage of superfamilies and folds. A total of 67% of SG domains in CATH are unique, lacking an already characterised close homologue in the PDB, whereas only 21% of non-SG domains are unique. For 29% of domains, structure determination revealed a remote evolutionary relationship not apparent from sequence, and 19% and 11% contributed new superfamilies and folds. The secondary structure class, fold and superfamily distributions of this dataset reflect those of the genomes. The domains fall into 172 different folds and 259 superfamilies in CATH but the distribution is highly skewed. The most populous of these are those that recur most frequently in the genomes. Whilst 11% of superfamilies are bacteria-specific, most are common to all three superkingdoms of life and together the 316 PDB entries have provided new and reliable homology models for 9287 non-redundant gene sequences in 206 completely sequenced genomes. From the perspective of this analysis, it appears that structural genomics is on track to be a success, and it is hoped that this work will inform future directions of the field.

Determining functional specificity from protein sequences

Given a large family of homologous protein sequences, many methods can divide the family into smaller groups that correspond to the different functions carried out by proteins within the family. One important problem, however, has been the absence of a general method for selecting an appropriate level of granularity, or size of the groups. We propose a consistent way of choosing the granularity that is independent of the sequence similarity and sequence clustering method used. We study three large, well-investigated protein families: basic leucine zippers, nuclear receptors and proteins with three consecutive C2H2 zinc fingers. Our method is tested against known functional information, the experimentally determined binding specificities, using a simple scoring method. The significance of the groups is also measured by randomizing the data. Finally, we compare our algorithm against a popular method of grouping proteins, the TRIBE-MCL method. In the end, we determine that dividing the families at the proposed level of granularity creates very significant and useful groups of proteins that correspond to the different DNA-binding motifs. We expect that such groupings will be useful in studying not only DNA binding but also other protein interactions.

Recapitulation of Protein Family Divergence using Flexible Backbone Protein Design

We use flexible backbone protein design to explore the sequence and structure neighborhoods of naturally occurring proteins. The method samples sequence and structure space in the vicinity of a known sequence and structure by alternately optimizing the sequence for a fixed protein backbone using rotamer based sequence search, and optimizing the backbone for a fixed amino acid sequence using atomic-resolution structure prediction. We find that such a flexible backbone design method better recapitulates protein family sequence variation than sequence optimization on fixed backbones or randomly perturbed backbone ensembles for ten diverse protein structures. For the SH3 domain, the backbone structure variation in the family is also better recapitulated than in randomly perturbed backbones. The potential application of this method as a model of protein family evolution is highlighted by a concerted transition to the amino acid sequence in the structural core of one SH3 domain starting from the backbone coordinates of an homologous structure.

Into the heart of darkness: large-scale clustering of human non-coding DNA.

It is currently believed that the human genome contains about twice as much non-coding functional regions as it does protein-coding genes, yet our understanding of these regions is very limited. We examine the intersection between syntenically conserved sequences in the human, mouse and rat genomes, and sequence similarities within the human genome itself, in search of families of non-protein-coding elements. For this purpose we develop a graph theoretic clustering algorithm, akin to the highly successful methods used in elucidating protein sequence family relationships. The algorithm is applied to a highly filtered set of about 700 000 human-rodent evolutionarily conserved regions, not resembling any known coding sequence, which encompasses 3.7% of the human genome. From these, we obtain roughly 12 000 non-singleton clusters, dense in significant sequence similarities. Further analysis of genomic location, evidence of transcription and RNA secondary structure reveals many clusters to be significantly homogeneous in one or more characteristics. This subset of the highly conserved non-protein-coding elements in the human genome thus contains rich family-like structures, which merit in-depth analysis. Supplementary material to this work is available at http://www.soe.ucsc.edu/~jill/dark.html

Particle Swarm Optimisation for Protein Motif Discovery

In this paper, a modified particle swarm optimisation algorithm is proposed for protein sequence motif discovery. Protein sequences are represented as a chain of symbols and a protein sequence motif is a short sequence that exists in most of the protein sequence families. Protein sequence symbols are converted into numbers using a one to one amino acid translation table. The simulation uses EGF protein and C2H2 Zinc Finger protein families obtained from the PROSITE database. Simulation results show that the modified particle swarm optimisation algorithm is effective in obtaining global optimum sequence patterns, achieving 96.9 and 99.5 classification accuracy respectively in EGF and C2H2 Zinc Finger protein families. A better true positive hit result is achieved when compared to the motifs published in PROSITE database.

FamClash: a method for ranking the activity of engineered enzymes.

This article introduces the computational procedure FamClash for analyzing incompatibilities in engineered protein hybrids by using protein family sequence data. All pairs of residue positions in the sequence alignment that conserve the property triplet of charge, volume, and hydrophobicity are first identified, and significant deviations are denoted as residue-residue clashes. This approach moves beyond earlier efforts aimed at solely classifying hybrids as functional or nonfunctional by correlating the rank ordering of these hybrids based on their activity levels. Experimental testing of this approach was performed in parallel to assess the predictive ability of FamClash. As a model system, single-crossover ITCHY (incremental truncation for the creation of hybrid enzymes) libraries were prepared from the Escherichia coli and Bacillus subtilis dihydrofolate reductases, and the activities of functional hybrids were determined. Comparisons of the predicted clash map as a function of crossover position revealed good agreement with activity data, reproducing the observed V shape and matching the location of a local peak in activity.