Inference of Protein Function from Protein Structure

Protein Science, 2006

We describe a method to assign a protein structure to a functional family using family-specific fingerprints. Fingerprints represent amino acid packing patterns that occur in most members of a family but are rare in the background, a nonredundant subset of PDB; their information is additional to sequence alignments, sequence patterns, structural superposition, and active-site templates. Fingerprints were derived for 120 families in SCOP using Frequent Subgraph Mining. For a new structure, all occurrences of these family-specific fingerprints may be found by a fast algorithm for subgraph isomorphism; the structure can then be assigned to a family with a confidence value derived from the number of fingerprints found and their distribution in background proteins. In validation experiments, we infer the function of new members added to SCOP families and we discriminate between structurally similar, but functionally divergent TIM barrel families. We then apply our method to predict function for several structural genomics proteins, including orphan structures. Some predictions have been corroborated by other computational methods and some validated by subsequent functional characterization.

Journal of Computer-Aided Molecular Design

Protein function prediction is one of the central problems in computational biology. We present a novel automated protein structure-based function prediction method using libraries of local residue packing patterns that are common to most proteins in a known functional family. Critical to this approach is the representation of a protein structure as a graph where residue vertices (residue name used as a vertex label) are connected by geometrical proximity edges. The approach employs two steps. First, it uses a fast subgraph mining algorithm to find all occurrences of family-specific labeled subgraphs for all well characterized protein structural and functional families. Second, it queries a new structure for occurrences of a set of motifs characteristic of a known family, using a graph index to speed up Ullman's subgraph isomorphism algorithm. The confidence of function inference from structure depends on the number of family-specific motifs found in the query structure compared...

Journal of Structural and Functional Genomics, 2011

The Protein Structure Initiative's Structural Biology Knowledgebase (SBKB, URL: http://sbkb.org) is an open web resource designed to turn the products of the structural genomics and structural biology efforts into knowledge that can be used by the biological community to understand living systems and disease. Here we will present examples on how to use the SBKB to enable biological research. For example, a protein sequence or Protein Data Bank (PDB) structure ID search will provide a list of related protein structures in the PDB, associated biological descriptions (annotations), homology models, structural genomics protein target status, experimental protocols, and the ability to order available DNA clones from the PSI:Biology-Materials Repository. A text search will find publication and technology reports resulting from the PSI's high-throughput research efforts. Web tools that aid in research, including a system that accepts protein structure requests from the community, will also be described. Created in collaboration with the Nature Publishing Group, the Structural Biology Knowledgebase monthly update also provides a research library, editorials about new research advances, news, and an events calendar to present a broader view of structural genomics and structural biology. Protein Model Portal Theoretical models http://www.proteinmodelportal.org PSI Technology Portal Technology reports http://technology.lbl.gov/portal/home PSI Publications Portal PSI publications http://olenka.med.virginia.edu/psi/ BioSync X-ray crystallography methods http://biosync.sbkb.org Community-Nominated Target proposal system Community requests for protein structures http://cnt.sbkb.org/CNT

Journal of Molecular Modeling, 2012

Enzootic pneumonia caused by Mycoplasma hyopneumoniae is a major constraint to efficient pork production throughout the world. This pathogen has a small genome with 716 coding sequences, of which 418 are homologous to proteins with known functions. However, almost 42% of the 716 coding sequences are annotated as hypothetical proteins. Alternative methodologies such as threading and comparative modeling can be used to predict structures and functions of such hypothetical proteins. Often, these alternative methods can answer questions about the properties of a model system faster than experiments. In this study, we predicted the structures of seven proteins annotated as hypothetical in M. hyopneumoniae, using the structure-based approaches mentioned above. Three proteins were predicted to be involved in metabolic processes, two proteins in transcription and two proteins where no function could be assigned. However, the modeled structures of the last two proteins suggested experimental designs to identify their functions. Our findings are important in diminishing the gap between the lack of annotation of important metabolic pathways and the great number of hypothetical proteins in the M. hyopneumoniae genome.

Nature Methods, 2013

As the number of sequenced genomes rapidly grows, the overwhelming majority of protein products can only be annotated computationally. if computational predictions are to be relied upon, it is crucial that the accuracy of these methods be high. here we report the results from the first large-scale community-based critical assessment of protein function annotation (CAFA) experiment. Fifty-four methods representing the state of the art for protein function prediction were evaluated on a target set of 866 proteins from 11 organisms. two findings stand out: (i) today's best protein function prediction algorithms substantially outperform widely used first-generation methods, with large gains on all types of targets; and (ii) although the top methods perform well enough to guide experiments, there is considerable need for improvement of currently available tools.

Proceedings of the National Academy of Sciences, 2011

BMC Bioinformatics, 2011

Background: In response to the rapid growth of available genome sequences, efforts have been made to develop automatic inference methods to functionally characterize them. Pipelines that infer functional annotation are now routinely used to produce new annotations at a genome scale and for a broad variety of species. These pipelines differ widely in their inference algorithms, confidence thresholds and data sources for reasoning. This heterogeneity makes a comparison of the relative merits of each approach extremely complex. The evaluation of the quality of the resultant annotations is also challenging given there is often no existing gold-standard against which to evaluate precision and recall.

BMC Systems Biology, 2013

Background: Construction of a reliable network remains the bottleneck for network-based protein function prediction. We built an artificial network model called protein overlap network (PON) for the entire genome of yeast, fly, worm, and human, respectively. Each node of the network represents a protein, and two proteins are connected if they share a domain according to InterPro database. Results: The function of a protein can be predicted by counting the occurrence frequency of GO (gene ontology) terms associated with domains of direct neighbors. The average success rate and coverage were 34.3% and 43.9%, respectively, for the test genomes, and were increased to 37.9% and 51.3% when a composite PON of the four species was used for the prediction. As a comparison, the success rate was 7.0% in the random control procedure. We also made predictions with GO term annotations of the second layer nodes using the composite network and obtained an impressive success rate (>30%) and coverage (>30%), even for small genomes. Further improvement was achieved by statistical analysis of manually annotated GO terms for each neighboring protein. Conclusions: The PONs are composed of dense modules accompanied by a few long distance connections. Based on the PONs, we developed multiple approaches effective for protein function prediction.

BMC Bioinformatics, 2013

Background: Many Automatic Function Prediction (AFP) methods were developed to cope with an increasing growth of the number of gene sequences that are available from high throughput sequencing experiments. To support the development of AFP methods, it is essential to have community wide experiments for evaluating performance of existing AFP methods. Critical Assessment of Function Annotation (CAFA) is one such community experiment. The meeting of CAFA was held as a Special Interest Group (SIG) meeting at the Intelligent Systems in Molecular Biology (ISMB) conference in 2011. Here, we perform a detailed analysis of two sequence-based function prediction methods, PFP and ESG, which were developed in our lab, using the predictions submitted to CAFA. Results: We evaluate PFP and ESG using four different measures in comparison with BLAST, Prior, and GOtcha. In addition to the predictions submitted to CAFA, we further investigate performance of a different scoring function to rank order predictions by PFP as well as PFP/ESG predictions enriched with Priors that simply adds frequently occurring Gene Ontology terms as a part of predictions. Prediction accuracies of each method were also evaluated separately for different functional categories. Successful and unsuccessful predictions by PFP and ESG are also discussed in comparison with BLAST. Conclusion: The in-depth analysis discussed here will complement the overall assessment by the CAFA organizers. Since PFP and ESG are based on sequence database search results, our analyses are not only useful for PFP and ESG users but will also shed light on the relationship of the sequence similarity space and functions that can be inferred from the sequences.

Journal of Structural and Functional Genomics, 2012

The structural genomics projects have been accumulating an increasing number of protein structures, many of which remain functionally unknown. In parallel effort to experimental methods, computational methods are expected to make a significant contribution for functional elucidation of such proteins. However, conventional computational methods that transfer functions from homologous proteins do not help much for these uncharacterized protein structures because they do not have apparent structural or sequence similarity with the known proteins. Here, we briefly review two avenues of computational function prediction methods, i.e. structure-based methods and sequence-based methods. The focus is on our recently developments of local structure-based methods and sequence-based methods, which can effectively extract function information from distantly related proteins. Two structure-based methods, Pocket-Surfer and Patch-Surfer, identify similar known ligand binding sites for pocket regions in a query protein without using global protein fold similarity information. Two sequence-based methods, PFP and ESG, make use of weakly similar sequences that are conventionally discarded in homology based function annotation. Combined together with experimental methods we hope that computational methods will make leading contribution in functional elucidation of the protein structures.

International Journal of Molecular Sciences, 2010

Functional elucidation of uncharacterized protein structures is an important task in bioinformatics. We report our new approach for structure-based function prediction which captures local surface features of ligand binding pockets. Function of proteins, specifically, binding ligands of proteins, can be predicted by finding similar local surface regions of known proteins. To enable partial comparison of binding sites in proteins, a weighted bipartite matching algorithm is used to match pairs of surface patches. The surface patches are encoded with the 3D Zernike descriptors. Unlike the existing methods which compare global characteristics of the protein fold or the global pocket shape, the local surface patch method can find functional similarity between non-homologous proteins and binding pockets for flexible ligand molecules. The proposed method improves prediction results over global pocket shape-based method which was previously developed by our group.

BMC Proceedings, 2012

Background: Advancements in function prediction algorithms are enabling large scale computational annotation for newly sequenced genomes. With the increase in the number of functionally well characterized proteins it has been observed that there are many proteins involved in more than one function. These proteins characterized as moonlighting proteins show varied functional behavior depending on the cell type, localization in the cell, oligomerization, multiple binding sites, etc. The functional diversity shown by moonlighting proteins may have significant impact on the traditional sequence based function prediction methods. Here we investigate how well diverse functions of moonlighting proteins can be predicted by some existing function prediction methods. Results: We have analyzed the performances of three major sequence based function prediction methods, PSI-BLAST, the Protein Function Prediction (PFP), and the Extended Similarity Group (ESG) on predicting diverse functions of moonlighting proteins. In predicting discrete functions of a set of 19 experimentally identified moonlighting proteins, PFP showed overall highest recall among the three methods. Although ESG showed the highest precision, its recall was lower than PSI-BLAST. Recall by PSI-BLAST greatly improved when BLOSUM45 was used instead of BLOSUM62. Conclusion: We have analyzed the performances of PFP, ESG, and PSI-BLAST in predicting the functional diversity of moonlighting proteins. PFP shows overall better performance in predicting diverse moonlighting functions as compared with PSI-BLAST and ESG. Recall by PSI-BLAST greatly improved when BLOSUM45 was used. This analysis indicates that considering weakly similar sequences in prediction enhances the performance of sequence based AFP methods in predicting functional diversity of moonlighting proteins. The current study will also motivate development of novel computational frameworks for automatic identification of such proteins.

BMC Bioinformatics, 2007

Background: Most methods for predicting functional sites in protein 3D structures, rely on information on related proteins and cannot be applied to proteins with no known relatives. Another limitation of these methods is the lack of a well annotated set of functional sites to use as benchmark for validating their predictions. Experimental findings and theoretical considerations suggest that residues involved in function often contribute unfavorably to the native state stability. We examine the possibility of systematically exploiting this intrinsic property to identify functional sites using an original procedure that detects destabilizing regions in protein structures. In addition, to relate destabilizing regions to known functional sites, a novel benchmark consisting of a diverse set of hand-curated protein functional sites is derived. Results: A procedure for detecting clusters of destabilizing residues in protein structures is presented. Individual residue contributions to protein stability are evaluated using detailed atomic models and a force-field successfully applied in computational protein design. The most destabilizing residues, and some of their closest neighbours, are clustered into destabilizing regions following a rigorous protocol. Our procedure is applied to high quality apo-structures of 63 unrelated proteins. The biologically relevant binding sites of these proteins were annotated using all available information, including structural data and literature curation, resulting in the largest hand-curated dataset of binding sites in proteins available to date. Comparing the destabilizing regions with the annotated binding sites in these proteins, we find that the overlap is on average limited, but significantly better than random. Results depend on the type of bound ligand. Significant overlap is obtained for most polysaccharide-and small ligand-binding sites, whereas no overlap is observed for most nucleic acid binding sites. These differences are rationalised in terms of the geometry and energetics of the binding site. Conclusion: We find that although destabilizing regions as detected here can in general not be used to predict binding sites in protein structures, they can provide useful information, particularly on the location of functional sites that bind polysaccharides and small ligands. This information can be exploited in methods for predicting function in protein structures with no known relatives. Our publicly available benchmark of hand-curated functional sites in proteins should help other workers derive and validate new prediction methods.

Scientific reports, 2017

Text mining has become an important tool in bioinformatics research with the massive growth in the biomedical literature over the past decade. Mining the biomedical literature has resulted in an incredible number of computational algorithms that assist many bioinformatics researchers. In this paper, we present a text mining system called Gene Interaction Rare Event Miner (GIREM) that constructs gene-gene-interaction networks for human genome using information extracted from biomedical literature. GIREM identifies functionally related genes based on their co-occurrences in the abstracts of biomedical literature. For a given gene g, GIREM first extracts the set of genes found within the abstracts of biomedical literature associated with g. GIREM aims at enhancing biological text mining approaches by identifying the semantic relationship between each co-occurrence of a pair of genes in abstracts using the syntactic structures of sentences and linguistics theories. It uses a supervised ...

BMC Bioinformatics, 2010

Background: There is an increasing number of proteins with known structure but unknown function. Determining their function would have a significant impact on understanding diseases and designing new therapeutics. However, experimental protein function determination is expensive and very time-consuming. Computational methods can facilitate function determination by identifying proteins that have high structural and chemical similarity. Results: We present LabelHash, a novel algorithm for matching substructural motifs to large collections of protein structures. The algorithm consists of two phases. In the first phase the proteins are preprocessed in a fashion that allows for instant lookup of partial matches to any motif. In the second phase, partial matches for a given motif are expanded to complete matches. The general applicability of the algorithm is demonstrated with three different case studies. First, we show that we can accurately identify members of the enolase superfamily with a single motif. Next, we demonstrate how LabelHash can complement SOIPPA, an algorithm for motif identification and pairwise substructure alignment. Finally, a large collection of Catalytic Site Atlas motifs is used to benchmark the performance of the algorithm. LabelHash runs very efficiently in parallel; matching a motif against all proteins in the 95% sequence identity filtered non-redundant Protein Data Bank typically takes no more than a few minutes. The LabelHash algorithm is available through a web server and as a suite of standalone programs at . The output of the LabelHash algorithm can be further analyzed with Chimera through a plugin that we developed for this purpose. Conclusions: LabelHash is an efficient, versatile algorithm for large-scale substructure matching. When LabelHash is running in parallel, motifs can typically be matched against the entire PDB on the order of minutes. The algorithm is able to identify functional homologs beyond the twilight zone of sequence identity and even beyond fold similarity. The three case studies presented in this paper illustrate the versatility of the algorithm.

BMC Bioinformatics, 2010

Background Structural variations caused by a wide range of physico-chemical and biological sources directly influence the function of a protein. For enzymatic proteins, the structure and chemistry of the catalytic binding site residues can be loosely defined as a substructure of the protein. Comparative analysis of drug-receptor substructures across and within species has been used for lead evaluation. Substructure-level similarity between the binding sites of functionally similar proteins has also been used to identify instances of convergent evolution among proteins. In functionally homologous protein families, shared chemistry and geometry at catalytic sites provide a common, local point of comparison among proteins that may differ significantly at the sequence, fold, or domain topology levels. Results This paper describes two key results that can be used separately or in combination for protein function analysis. The Family-wise Analysis of SubStructural Templates (FASST) method...

Proteins: Structure, Function, and Bioinformatics, 2011

Understanding protein function is one of the keys to understanding life at the molecular level. It is also important in the context of human disease because many conditions arise as a consequence of alterations of protein function. The recent availability

BMC Bioinformatics, 2007

Background: Genome-wide association studies of common diseases for common, low penetrance causal variants are underway. A proportion of these will alter protein sequences, the most common of which is the non-synonymous single nucleotide polymorphism (nsSNP). It would be an advantage if the functional effects of an nsSNP on protein structure and function could be predicted, both for the final identification process of a causal variant in a disease-associated chromosome region, and in further functional analyses of the nsSNP and its disease-associated protein. Results: In the present report we have compared and contrasted structure-and sequence-based methods of prediction to over 5500 genes carrying nearly 24,000 nsSNPs, by employing an automatic comparative modelling procedure to build models for the genes. The nsSNP information came from two sources, the OMIM database which are rare (minor allele frequency, MAF, < 0.01) and are known to cause penetrant, monogenic diseases. Secondly, nsSNP information came from dbSNP125, for which the vast majority of nsSNPs, mostly MAF > 0.05, have no known link to a disease. For over 40% of the nsSNPs, structure-based methods predicted which of these sequence changes are likely to either disrupt the structure of the protein or interfere with the function or interactions of the protein. For the remaining 60%, we generated sequence-based predictions. Conclusion: We show that, in general, the prediction tools are able distinguish disease causing mutations from those mutations which are thought to have a neutral affect. We give examples of mutations in genes that are predicted to be deleterious and may have a role in disease. Contrary to previous reports, we also show that rare mutations are consistently predicted to be deleterious as often as commonly occurring nsSNPs.

Computers in Biology and Medicine, 2021

Mutations in proto-oncogenes (OG) and the loss of regulatory function of tumor suppression genes (TSG) are the common underlying mechanism for uncontrolled tumor growth. While cancer is a heterogeneous complex of distinct diseases, finding the potentiality of the genes related functionality to OG or TSG through computational studies can help develop drugs that target the disease. This paper proposes a classification method that starts with a preprocessing stage to extract the feature map sets from the input 3D protein structural information. The next stage is a deep convolutional neural network stage (DCNN) that outputs the probability of functional classification of genes. We explored and tested two approaches: in Approach 1, all filtered and cleaned 3D-protein-structures (PDB) are pooled together, whereas in Approach 2, the primary structures and their corresponding PDBs are separated according to the genes' primary structural information. Following the DCNN stage, a dynamic programming-based method is used to determine the final prediction of the primary structures' functionality. We validated our proposed method using the COSMIC online database. For the OG vs TSC classification problem the AUROC of the DCNN stage for Approach 1 and Approach 2 DCNN are 0.978 and 0.765, respectively. The AUROCs of the final genes' primary structure functionality classification for Approach 1 and Approach 2 are 0.989, and 0.879, respectively. For comparison, the current state-of-the-art reported AUROC is 0.924. Our results warrant further study to apply the deep learning models to humans' (GRCh38) genes, for predicting their corresponding probabilities of functionality in the cancer drivers.

Scientific Reports, 2018

Lentinula edodes is a popular cultivated edible mushroom with high nutritional and medicinal value. To understand the regulation of gene expression in the dikaryotic mycelium and mature fruiting body in the commercially important Korean L. edodes strain, we first performed comparative transcriptomic analysis, using Illumina HiSeq platform. De novo assembly of these sequences revealed 11,675 representative transcripts in two different stages of L. edodes. A total of 9,092 unigenes were annotated and subjected to Gene Ontology, EuKaryotic Orthologous Groups, and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses. Gene expression analysis revealed that 2,080 genes were differentially expressed, with 1,503 and 577 upregulated in the mycelium and a mature fruiting body, respectively. Analysis of 18 KEGG categories indicated that fruiting body-specific transcripts were significantly enriched in ‘replication and repair’ and ‘transcription’ pathways, which are important for premeiotic ...

Journal of Grid Computing, 2018

Accurate annotation of protein functions is important for a profound understanding of molecular biology. A large number of proteins remain uncharacterized because of the sparsity of available supporting information. For a large set of uncharacterized proteins, the only type of information available is their amino acid sequence. In this paper, we propose DeepSeq -a deep learning architecture -that utilizes only the protein sequence information to predict its associated functions. The prediction process does not require handcrafted features; rather, the architecture automatically extracts representations from the input sequence data. Results of our experiments with DeepSeq indicate significant improvements in terms of prediction accuracy when compared with other sequence-based methods. Our deep learning model achieves an overall validation accuracy of 86.72%, with an F1 score of 71.13%. Moreover, using the automatically learned features and without any changes to DeepSeq, we successfully solved a different problem i.e. protein function localization, with no human intervention. Finally, we discuss how this same architecture can be used to solve even more complicated problems such as prediction of 2D and 3D structure as well as protein-protein interactions.