Automated structure-based prediction of functional sites in proteins: applications to assessing the validity of inheriting protein function from homology in genome annotation and to protein docking

A major problem in genome annotation is whether it is valid to transfer the function from a characterised protein to a homologue of unknown activity. Here, we show that one can employ a strategy that uses a structure-based prediction of protein functional sites to assess the reliability of functional inheritance. We have automated and benchmarked a method based on the evolutionary trace approach. Using a multiple sequence alignment, we identified invariant polar residues, which were then mapped onto the protein structure. Spatial clusters of these invariant residues formed the predicted functional site. For 68 of 86 proteins examined, the method yielded information about the observed functional site. This algorithm for functional site prediction was then used to assess the validity of transferring the function between homologues. This procedure was tested on 18 pairs of homologous proteins with unrelated function and 70 pairs of proteins with related function, and was shown to be 94 % accurate. This automated method could be linked to schemes for genome annotation. Finally, we examined the use of functional site prediction in protein-protein and protein-DNA docking. The use of predicted functional sites was shown to filter putative docked complexes with a discrimination similar to that obtained by manually including biological information about active sites or DNA-binding residues.     

Functional annotation of proteomic sequences based on consensus of sequence and structural analysis

To maximise the assignment of function of the proteins encoded by a genome and to aid the search for novel drug targets, there is an emerging need for sensitive methods of predicting protein function on a genome-wide basis. GeneAtlas is an automated, high-throughput pipeline for the prediction of protein structure and function using sequence similarity detection, homology modelling and fold recognition methods. GeneAtlas is described in detail here. To test GeneAtlas, a 'virtual' genome was used, a subset of PDB structures from the SCOP database, in which the functional relationships are known. GeneAtlas detects additional relationships by building 3D models in comparison with the sequence searching method PSI-BLAST. Functionally related proteins with sequence identity below the twilight zone can be recognised correctly.     

Identification of novel families of membrane proteins from the model plant Arabidopsis thaliana

MOTIVATION: The completion of the Arabidopsis genome offers the first opportunity to analyze all of the membrane protein sequences of a plant. The majority of integral membrane proteins including transporters, channels, and pumps contain hydrophobic alpha-helices and can be selected based on TransMembrane Spanning (TMS) domain prediction. By clustering the predicted membrane proteins based on sequence, it is possible to sort the membrane proteins into families of known function, based on experimental evidence or homology, or unknown function. This provides a way to identify target sequences for future functional analysis. RESULTS: An automated approach was used to select potential membrane protein sequences from the set of all predicted proteins and cluster the sequences into related families. The recently completed sequence of Arabidopsis thaliana, a model plant, was analyzed. Of the 25,470 predicted protein sequences 4589 (18%) were identified as containing two or more membrane spanning domains. The membrane protein sequences clustered into 628 distinct families containing 3208 sequences. Of these, 211 families (1764 sequences) either contained proteins of known function or showed homology to proteins of known function in other species. However, 417 families (1444 sequences) contained only sequences with no known function and no homology to proteins of known function. In addition, 1381 sequences did not cluster with any family and no function could be assigned to 1337 of these.     

In Silico Characterization of Proteins: UniProt,InterPro and Integr8

Nucleic acid sequences from genome sequencing projects are submitted as raw data, from which biologists attempt to elucidate the function of the predicted gene products. The protein sequences are stored in public databases, such as the UniProt Knowledgebase (UniProtKB), where curators try to add predicted and experimental functional information. Protein function prediction can be done using sequence similarity searches, but an alternative approach is to use protein signatures, which classify proteins into families and domains. The major protein signature databases are available through the integrated InterPro database, which provides a classification of UniProtKB sequences. As well as characterization of proteins through protein families, many researchers are interested in analyzing the complete set of proteins from a genome (i.e. the proteome), and there are databases and resources that provide non-redundant proteome sets and analyses of proteins from organisms with completely sequenced genomes. This article reviews the tools and resources available on the web for single and large-scale protein characterization and whole proteome analysis.     

An efficient algorithm for large-scale detection of protein families

Detection of protein families in large databases is one of the principal research objectives in structural and functional genomics. Protein family classification can significantly contribute to the delineation of functional diversity of homologous proteins, the prediction of function based on domain architecture or the presence of sequence motifs as well as comparative genomics, providing valuable evolutionary insights. We present a novel approach called TRIBE-MCL for rapid and accurate clustering of protein sequences into families. The method relies on the Markov cluster (MCL) algorithm for the assignment of proteins into families based on precomputed sequence similarity information. This novel approach does not suffer from the problems that normally hinder other protein sequence clustering algorithms, such as the presence of multi-domain proteins, promiscuous domains and fragmented proteins. The method has been rigorously tested and validated on a number of very large databases, including SwissProt, InterPro, SCOP and the draft human genome. Our results indicate that the method is ideally suited to the rapid and accurate detection of protein families on a large scale. The method has been used to detect and categorise protein families within the draft human genome and the resulting families have been used to annotate a large proportion of human proteins.     

From patenting genes to proteins: the search for utility via function

The debate regarding the patenting of genes has extended into the post-genome era. With only approximately 35000 genes deduced from the draft sequence of the human genome, there are fears that a few companies have already gained monopoly on the potential benefits from this knowledge. Nevertheless, it is accepted that proteins determine gene function and function is not readily predicted from gene sequence. Furthermore, genes can encode multiple proteins and a single protein can have multiple functions. Here, we argue that unraveling the intrinsic complexity of proteins and their functions is the key towards determining the utility requirement for patenting protein inventions and consider the possible socioeconomic impact.     

Functional insights from structural predictions: analysis of the Escherichia coli genome

Fold assignments for proteins from the Escherichia coli genome are carried out using BASIC, a profile-profile alignment algorithm, recently tested on fold recognition benchmarks and on the Mycoplasma genitalium genome and PSI BLAST, the newest generation of the de facto standard in homology search algorithms. The fold assignments are followed by automated modeling and the resulting three-dimensional models are analyzed for possible function prediction. Close to 30% of the proteins encoded in the E. coli genome can be recognized as homologous to a protein family with known structure. Most of these homologies (23% of the entire genome) can be recognized both by PSI BLAST and BASIC algorithms, but the latter recognizes an additional 260 homologies. Previous estimates suggested that only 10-15% of E. coli proteins can be characterized this way. This dramatic increase in the number of recognized homologies between E. coli proteins and structurally characterized protein families is partly due to the rapid increase of the database of known protein structures, but mostly it is due to the significant improvement in prediction algorithms. Knowing protein structure adds a new dimension to our understanding of its function and the predictions presented here can be used to predict function for uncharacterized proteins. Several examples, analyzed in more detail in this paper, include the DPS protein protecting DNA from oxidative damage (predicted to be homologous to ferritin with iron ion acting as a reducing agent) and the ahpC/tsa family of proteins, which provides resistance to various oxidating agents (predicted to be homologous to glutathione peroxidase).     

Prediction of partial membrane protein topologies using a consensus approach

We have developed a method to reliably identify partial membrane protein topologies using the consensus of five topology prediction methods. When evaluated on a test set of experimentally characterized proteins, we find that approximately 90% of the partial consensus topologies are correctly predicted in membrane proteins from prokaryotic as well as eukaryotic organisms. Whole-genome analysis reveals that a reliable partial consensus topology can be predicted for approximately 70% of all membrane proteins in a typical bacterial genome and for approximately 55% of all membrane proteins in a typical eukaryotic genome. The average fraction of sequence length covered by a partial consensus topology is 44% for the prokaryotic proteins and 17% for the eukaryotic proteins in our test set, and similar numbers are found when the algorithm is applied to whole genomes. Reliably predicted partial topologies may simplify experimental determinations of membrane protein topology.     

Computational differentiation of N-terminal signal peptides and transmembrane helices

Signal peptides and transmembrane helices both contain a stretch of hydrophobic amino acids. This common feature makes it difficult for signal peptide and transmembrane helix predictors to correctly assign identity to stretches of hydrophobic residues near the N-terminal methionine of a protein sequence. The inability to reliably distinguish between N-terminal transmembrane helix and signal peptide is an error with serious consequences for the prediction of protein secretory status or transmembrane topology. In this study, we report a new method for differentiating protein N-terminal signal peptides and transmembrane helices. Based on the sequence features extracted from hydrophobic regions (amino acid frequency, hydrophobicity, and the start position), we set up discriminant functions and examined them on non-redundant datasets with jackknife tests. This method can incorporate other signal peptide prediction methods and achieve higher prediction accuracy. For Gram-negative bacterial proteins, 95.7% of N-terminal signal peptides and transmembrane helices can be correctly predicted (coefficient 0.90). Given a sensitivity of 90%, transmembrane helices can be identified from signal peptides with a precision of 99% (coefficient 0.92). For eukaryotic proteins, 94.2% of N-terminal signal peptides and transmembrane helices can be correctly predicted with coefficient 0.83. Given a sensitivity of 90%, transmembrane helices can be identified from signal peptides with a precision of 87% (coefficient 0.85). The method can be used to complement current transmembrane protein prediction and signal peptide prediction methods to improve their prediction accuracies.     

Meta-basic estimates the size of druggable human genome

We present here the estimation of the upper limit of the number of molecular targets in the human genome that represent an opportunity for further therapeutic treatment. We select around ∼6300 human proteins that are similar to sequences of known protein targets collected from DrugBank database. Our bioinformatics study estimates the size of ‘druggable’ human genome to be around 20% of human proteome, i.e. the number of the possible protein targets for small-molecule drug design in medicinal chemistry. We do not take into account any toxicity prediction, the three-dimensional characteristics of the active site in the predicted ‘druggable’ protein families, or detailed chemical analysis of known inhibitors/drugs. Instead we rely on remote homology detection method Meta-BASIC, which is based on sequence and structural similarity. The prepared dataset of all predicted protein targets from human genome presents the unique opportunity for developing and benchmarking various in silico chemo/bio-informatics methods in the context of the virtual high throughput screening.     

GenTHREADER: an efficient and reliable protein fold recognition method for genomic sequences

A new protein fold recognition method is described which is both fast and reliable. The method uses a traditional sequence alignment algorithm to generate alignments which are then evaluated by a method derived from threading techniques. As a final step, each threaded model is evaluated by a neural network in order to produce a single measure of confidence in the proposed prediction. The speed of the method, along with its sensitivity and very low false-positive rate makes it ideal for automatically predicting the structure of all the proteins in a translated bacterial genome (proteome). The method has been applied to the genome of Mycoplasma genitalium, and analysis of the results shows that as many as 46 % of the proteins derived from the predicted protein coding regions have a significant relationship to a protein of known structure. In some cases, however, only one domain of the protein can be predicted, giving a total coverage of 30 % when calculated as a fraction of the number of amino acid residues in the whole proteome.     

Distinguishing enzyme structures from non-enzymes without alignments

The ability to predict protein function from structure is becoming increasingly important as the number of structures resolved is growing more rapidly than our capacity to study function. Current methods for predicting protein function are mostly reliant on identifying a similar protein of known function. For proteins that are highly dissimilar or are only similar to proteins also lacking functional annotations, these methods fail. Here, we show that protein function can be predicted as enzymatic or not without resorting to alignments. We describe 1178 high-resolution proteins in a structurally non-redundant subset of the Protein Data Bank using simple features such as secondary-structure content, amino acid propensities, surface properties and ligands. The subset is split into two functional groupings, enzymes and non-enzymes. We use the support vector machine-learning algorithm to develop models that are capable of assigning the protein class. Validation of the method shows that the function can be predicted to an accuracy of 77% using 52 features to describe each protein. An adaptive search of possible subsets of features produces a simplified model based on 36 features that predicts at an accuracy of 80%. We compare the method to sequence-based methods that also avoid calculating alignments and predict a recently released set of unrelated proteins. The most useful features for distinguishing enzymes from non-enzymes are secondary-structure content, amino acid frequencies, number of disulphide bonds and size of the largest cleft. This method is applicable to any structure as it does not require the identification of sequence or structural similarity to a protein of known function.     

Gene structure prediction by spliced alignment of genomic DNA with protein sequences: increased accuracy by differential splice site scoring

Gene identification in genomic DNA from eukaryotes is complicated by the vast combinatorial possibilities of potential exon assemblies. If the gene encodes a protein that is closely related to known proteins, gene identification is aided by matching similarity of potential translation products to those target proteins. The genomic DNA and protein sequences can be aligned directly by scoring the implied residues of in-frame nucleotide triplets against the protein residues in conventional ways, while allowing for long gaps in the alignment corresponding to introns in the genomic DNA. We describe a novel method for such spliced alignment. The method derives an optimal alignment based on scoring for both sequence similarity of the predicted gene product to the protein sequence and intrinsic splice site strength of the predicted introns. Application of the method to a representative set of 50 known genes from Arabidopsis thaliana showed significant improvement in prediction accuracy compared to previous spliced alignment methods. The method is also more accurate than ab initio gene prediction methods, provided sufficiently close target proteins are available. In view of the fast growth of public sequence repositories, we argue that close targets will be available for the majority of novel genes, making spliced alignment an excellent practical tool for high-throughput automated genome annotation.     

SECOM: a novel hash seed and community detection based-approach for genome-scale protein domain identification

With rapid advances in the development of DNA sequencing technologies, a plethora of high-throughput genome and proteome data from a diverse spectrum of organisms have been generated. The functional annotation and evolutionary history of proteins are usually inferred from domains predicted from the genome sequences. Traditional database-based domain prediction methods cannot identify novel domains, however, and alignment-based methods, which look for recurring segments in the proteome, are computationally demanding. Here, we propose a novel genome-wide domain prediction method, SECOM. Instead of conducting all-against-all sequence alignment, SECOM first indexes all the proteins in the genome by using a hash seed function. Local similarity can thus be detected and encoded into a graph structure, in which each node represents a protein sequence and each edge weight represents the shared hash seeds between the two nodes. SECOM then formulates the domain prediction problem as an overlapping community-finding problem in this graph. A backward graph percolation algorithm that efficiently identifies the domains is proposed. We tested SECOM on five recently sequenced genomes of aquatic animals. Our tests demonstrated that SECOM was able to identify most of the known domains identified by InterProScan. When compared with the alignment-based method, SECOM showed higher sensitivity in detecting putative novel domains, while it was also three orders of magnitude faster. For example, SECOM was able to predict a novel sponge-specific domain in nucleoside-triphosphatase (NTPases). Furthermore, SECOM discovered two novel domains, likely of bacterial origin, that are taxonomically restricted to sea anemone and hydra. SECOM is an open-source program and available at http://sfb.kaust.edu.sa/Pages/Software.aspx.     

Prediction of human cDNA from its homologous mouse full-length cDNA and human shotgun database

We propose a prediction method for human full-length cDNA by comparing sequence data between human genome shotgun sequence and mouse full-length cDNA. The human genome which is homologous to the mouse full-length cDNA is selected by a homology search program, and the predicted exons are connected at the exon-intron junction which gives the best homology score to the mouse full-length cDNA. The accuracy of the predicted human full-length coding region is 83.3%, and the false positive rate is 16.7%. Five human full-length proteins out of 20 proteins are correctly predicted.     

RBPBind: Quantitative Prediction of Protein-RNA Interactions

There are hundreds of RNA binding proteins in the human genome alone and their interactions with messenger and other RNAs in a cell regulate every step in an RNA's life cycle. To understand this interplay of proteins and RNA it is important to be able to know which protein binds which RNA how strongly and where. Here, we introduce RBPBind, a web-based tool for the quantitative prediction of the interaction of single-stranded RNA binding proteins with target RNAs that fully takes into account the effect of RNA secondary structure on binding affinity. Given a user-specified RNA and a protein selected from a set of several RNA-binding proteins, RBPBind computes their binding curve and effective binding constant. The server also computes the probability that, at a given protein concentration, a protein molecule will bind to any particular nucleotide along the RNA. The sequence specificity of the protein-RNA interaction is parameterized from public RNAcompete experiments and integrated into the recursions of the Vienna RNA package to simultaneously take into account protein binding and RNA secondary structure. We validate our approach by comparison to experimentally determined binding affinities of the HuR protein for several RNAs of different sequence contexts from the literature, showing that integration of raw sequence affinities into RNA secondary structure prediction significantly improves the agreement between computationally predicted and experimentally measured binding affinities. Our resource thus provides a quick and easy way to obtain reliable predicted binding affinities and locations for single-stranded RNA binding proteins based on RNA sequence alone.