SPINE-D: accurate prediction of short and long disordered regions by a single neural-network based method

Short and long disordered regions of proteins have different preference for different amino acid residues. Different methods often have to be trained to predict them separately. In this study, we developed a single neural-network-based technique called SPINE-D that makes a three-state prediction first (ordered residues and disordered residues in short and long disordered regions) and reduces it into a two-state prediction afterwards. SPINE-D was tested on various sets composed of different combinations of Disprot annotated proteins and proteins directly from the PDB annotated for disorder by missing coordinates in X-ray determined structures. While disorder annotations are different according to Disprot and X-ray approaches, SPINE-D's prediction accuracy and ability to predict disorder are relatively independent of how the method was trained and what type of annotation was employed but strongly depend on the balance in the relative populations of ordered and disordered residues in short and long disordered regions in the test set. With greater than 85% overall specificity for detecting residues in both short and long disordered regions, the residues in long disordered regions are easier to predict at 81% sensitivity in a balanced test dataset with 56.5% ordered residues but more challenging (at 65% sensitivity) in a test dataset with 90% ordered residues. Compared to eleven other methods, SPINE-D yields the highest area under the curve (AUC), the highest Mathews correlation coefficient for residue-based prediction, and the lowest mean square error in predicting disorder contents of proteins for an independent test set with 329 proteins. In particular, SPINE-D is comparable to a meta predictor in predicting disordered residues in long disordered regions and superior in short disordered regions. SPINE-D participated in CASP 9 blind prediction and is one of the top servers according to the official ranking. In addition, SPINE-D was examined for prediction of functional molecular recognition motifs in several case studies.     

SPINE-D: Accurate Prediction of Short and Long Disordered Regions by a Single Neural-Network Based Method

Abstract

Short and long disordered regions of proteins have different preference for different amino acid residues. Different methods often have to be trained to predict them separately. In this study, we developed a single neural-network-based technique called SPINE-D that makes a three-state prediction first (ordered residues and disordered residues in short and long disordered regions) and reduces it into a two-state prediction afterwards. SPINE-D was tested on various sets composed of different combinations of Disprot annotated proteins and proteins directly from the PDB annotated for disorder by missing coordinates in X-ray determined structures. While disorder annotations are different according to Disprot and X-ray approaches, SPINE-D's prediction accuracy and ability to predict disorder are relatively independent of how the method was trained and what type of annotation was employed but strongly depend on the balance in the relative populations of ordered and disordered residues in short and long disordered regions in the test set. With greater than 85% overall specificity for detecting residues in both short and long disordered regions, the residues in long disordered regions are easier to predict at 81% sensitivity in a balanced test dataset with 56.5% ordered residues but more challenging (at 65% sensitivity) in a test dataset with 90% ordered residues. Compared to eleven other methods, SPINE-D yields the highest area under the curve (AUC), the highest Mathews correlation coefficient for residue-based prediction, and the lowest mean square error in predicting disorder contents of proteins for an independent test set with 329 proteins. In particular, SPINE-D is comparable to a meta predictor in predicting disordered residues in long disordered regions and superior in short disordered regions. SPINE-D participated in CASP 9 blind prediction and is one of the top servers according to the official ranking. In addition, SPINE-D was examined for prediction of functional molecular recognition motifs in several case studies. The server and databases are available at http://sparks.informatics.iupui.edu/.     

Natively Disordered Proteins

Proteins can exist in at least three forms: the ordered form (solid-like), the partially folded form (collapsed, molten globule-like or liquid-like) and the extended form (extended, random coil-like or gas-like). The protein trinity hypothesis has two components: (i) a given native protein can be in any one of the three forms, depending on the sequence and the environment; and (ii) function can arise from any one of the three forms or from transitions between them. In this study, bioinformatics and data mining were used to investigate intrinsic disorder in proteins and develop neural network-based predictors of natural disordered regions (PONDR) that can discriminate between ordered and disordered residues with up to 84% accuracy. Predictions of intrinsic disorder indicate that the three kingdoms follow the disorder ranking eubacteria < archaebacteria < eukaryotes, with approximately half of eukaryotic proteins predicted to contain substantial regions of intrinsic disorder. Many of the known disordered regions are involved in signalling, regulation or control. Involvement of highly flexible or disordered regions in signalling is logical: a flexible sensor more readily undergoes conformational change in response to environmental perturbations than does a rigid one. Thus, the increased disorder in the eukaryotes is likely the direct result of an increased need for signalling and regulation in nucleated organisms. PONDR can also be used to detect molecular recognition elements that are disordered in the unbound state and become structured when bound to a biologically meaningful partner. Application of disorder predictions to cell-signalling, cancer-associated and control protein databases supports the widespread occurrence of protein disorder in these processes.     

Intrinsic disorder in the Protein Data Bank

The Protein Data Bank (PDB) is the preeminent source of protein structural information. PDB contains over 32,500 experimentally determined 3-D structures solved using X-ray crystallography or nuclear magnetic resonance spectroscopy. Intrinsically disordered regions fail to form a fixed 3-D structure under physiological conditions. In this study, we compare the amino-acid sequences of proteins whose structures are determined by X-ray crystallography with the corresponding sequences from the Swiss-Prot database. The analyzed dataset includes 16,370 structures, which represent 18,101 PDB chains and 5,434 different proteins from 910 different organisms (2,793 eukaryotic, 2,109 bacterial, 288 viral, and 244 archaeal). In this dataset, on average, each Swiss-Prot protein is represented by 7 PDB chains with 76% of the crystallized regions being represented by more than one structure. Intriguingly, the complete sequences of only approximately 7% of proteins are observed in the corresponding PDB structures, and only approximately 25% of the total dataset have >95% of their lengths observed in the corresponding PDB structures. This suggests that the vast majority of PDB proteins is shorter than their corresponding Swiss-Prot sequences and/or contain numerous residues, which are not observed in maps of electron density. To determine the prevalence of disordered regions in PDB, the residues in the Swiss-Prot sequences were grouped into four general categories, "Observed" (which correspond to structured regions), "Not observed" (regions with missing electron density, potentially disordered), "Uncharacterized," and "Ambiguous," depending on their appearance in the corresponding PDB entries. This non-redundant set of residues can be viewed as a 'fragment' or empirical domain database that contains a set of experimentally determined structured regions or domains and a set of experimentally verified disordered regions or domains. We studied the propensities and properties of residues in these four categories and analyzed their relations to the predictions of disorder using several algorithms. "Non-observed," "Ambiguous," and "Uncharacterized" regions were shown to possess the amino acid compositional biases typical of intrinsically disordered proteins. The application of four different disorder predictors (PONDR(R) VL-XT, VL3-BA, VSL1P, and IUPred) revealed that the vast majority of residues in the "Observed" dataset are ordered, and that the "Not observed" regions are mostly disordered. The "Uncharacterized" regions possess some tendency toward order, whereas the predictions for the short "Ambiguous" regions are really ambiguous. Long "Ambiguous" regions (>70 amino acid residues) are mostly predicted to be ordered, suggesting that they are likely to be "wobbly" domains. Overall, we showed that completely ordered proteins are not highly abundant in PDB and many PDB sequences have disordered regions. In fact, in the analyzed dataset approximately 10% of the PDB proteins contain regions of consecutive missing or ambiguous residues longer than 30 amino-acids and approximately 40% of the proteins possess short regions (> or =10 and < 30 amino-acid long) of missing and ambiguous residues.     

Intrinsic Disorder in the Protein Data Bank

Abstract

The Protein Data Bank (PDB) is the preeminent source of protein structural information. PDB contains over 32,500 experimentally determined 3-D structures solved using X-ray crystallography or nuclear magnetic resonance spectroscopy. Intrinsically disordered regions fail to form a fixed 3-D structure under physiological conditions. In this study, we compare the amino-acid sequences of proteins whose structures are determined by X-ray crystallography with the corresponding sequences from the Swiss-Prot database. The analyzed dataset includes 16,370 structures, which represent 18,101 PDB chains and 5,434 different proteins from 910 different organisms (2,793 eukaryotic, 2,109 bacterial, 288 viral, and 244 archaeal). In this dataset, on average, each Swiss-Prot protein is represented by 7 PDB chains with 76% of the crystallized regions being represented by more than one structure. Intriguingly, the complete sequences of only ~7% of proteins are observed in the corresponding PDB structures, and only ~25% of the total dataset have >95% of their lengths observed in the corresponding PDB structures. This suggests that the vast majority of PDB proteins is shorter than their corresponding Swiss-Prot sequences and/or contain numerous residues, which are not observed in maps of electron density. To determine the prevalence of disordered regions in PDB, the residues in the Swiss-Prot sequences were grouped into four general categories, “Observed” (which correspond to structured regions), “Not observed” (regions with missing electron density, potentially disordered), “Uncharacterized,” and “Ambiguous,” depending on their appearance in the corresponding PDB entries. This non-redundant set of residues can be viewed as a ‘fragment’ or empirical domain database that contains a set of experimentally determined structured regions or domains and a set of experimentally verified disordered regions or domains. We studied the propensities and properties of residues in these four categories and analyzed their relations to the predictions of disorder using several algorithms. “Non-observed,” “Ambiguous,” and “Uncharacterized” regions were shown to possess the amino acid compositional biases typical of intrinsically disordered proteins. The application of four different disorder predictors (PONDR® VL-XT, VL3-BA, VSL1P, and IUPred) revealed that the vast majority of residues in the “Observed” dataset are ordered, and that the “Not observed” regions are mostly disordered. The “Uncharacterized” regions possess some tendency toward order, whereas the predictions for the short “Ambiguous” regions are really ambiguous. Long “Ambiguous” regions (>70 amino acid residues) are mostly predicted to be ordered, suggesting that they are likely to be “wobbly” domains.

Overall, we showed that completely ordered proteins are not highly abundant in PDB and many PDB sequences have disordered regions. In fact, in the analyzed dataset ~10% of the PDB proteins contain regions of consecutive missing or ambiguous residues longer than 30 amino-acids and ~40% of the proteins possess short regions (≥10 and <30 amino-acid long) of missing and ambiguous residues.     

The importance of intrinsic disorder for protein phosphorylation

Reversible protein phosphorylation provides a major regulatory mechanism in eukaryotic cells. Due to the high variability of amino acid residues flanking a relatively limited number of experimentally identified phosphorylation sites, reliable prediction of such sites still remains an important issue. Here we report the development of a new web-based tool for the prediction of protein phosphorylation sites, DISPHOS (DISorder-enhanced PHOSphorylation predictor, http://www.ist.temple. edu/DISPHOS). We observed that amino acid compositions, sequence complexity, hydrophobicity, charge and other sequence attributes of regions adjacent to phosphorylation sites are very similar to those of intrinsically disordered protein regions. Thus, DISPHOS uses position-specific amino acid frequencies and disorder information to improve the discrimination between phosphorylation and non-phosphorylation sites. Based on the estimates of phosphorylation rates in various protein categories, the outputs of DISPHOS are adjusted in order to reduce the total number of misclassified residues. When tested on an equal number of phosphorylated and non-phosphorylated residues, the accuracy of DISPHOS reaches 76% for serine, 81% for threonine and 83% for tyrosine. The significant enrichment in disorder-promoting residues surrounding phosphorylation sites together with the results obtained by applying DISPHOS to various protein functional classes and proteomes, provide strong support for the hypothesis that protein phosphorylation predominantly occurs within intrinsically disordered protein regions.     

Evolutionary rate heterogeneity in proteins with long disordered regions

The dominant view in protein science is that a three-dimensional (3-D) structure is a prerequisite for protein function. In contrast to this dominant view, there are many counterexample proteins that fail to fold into a 3-D structure, or that have local regions that fail to fold, and yet carry out function. Protein without fixed 3-D structure is called intrinsically disordered. Motivated by anecdotal accounts of higher rates of sequence evolution in disordered protein than in ordered protein we are exploring the molecular evolution of disordered proteins. To test whether disordered protein evolves more rapidly than ordered protein, pairwise genetic distances were compared between the ordered and the disordered regions of 26 protein families having at least one member with a structurally characterized region of disorder of 30 or more consecutive residues. For five families, there were no significant differences in pairwise genetic distances between ordered and disordered sequences. The disordered region evolved significantly more rapidly than the ordered region for 19 of the 26 families. The functions of these disordered regions are diverse, including binding sites for protein, DNA, or RNA and also including flexible linkers. The functions of some of these regions are unknown. The disordered regions evolved significantly more slowly than the ordered regions for the two remaining families. The functions of these more slowly evolving disordered regions include sites for DNA binding. More work is needed to understand the underlying causes of the variability in the evolutionary rates of intrinsically ordered and disordered protein.     

Markov models of amino acid substitution to study proteins with intrinsically disordered regions

Background

Intrinsically disordered proteins (IDPs) or proteins with disordered regions (IDRs) do not have a well-defined tertiary structure, but perform a multitude of functions, often relying on their native disorder to achieve the binding flexibility through changing to alternative conformations. Intrinsic disorder is frequently found in all three kingdoms of life, and may occur in short stretches or span whole proteins. To date most studies contrasting the differences between ordered and disordered proteins focused on simple summary statistics. Here, we propose an evolutionary approach to study IDPs, and contrast patterns specific to ordered protein regions and the corresponding IDRs.

Results

Two empirical Markov models of amino acid substitutions were estimated, based on a large set of multiple sequence alignments with experimentally verified annotations of disordered regions from the DisProt database of IDPs. We applied new methods to detect differences in Markovian evolution and evolutionary rates between IDRs and the corresponding ordered protein regions. Further, we investigated the distribution of IDPs among functional categories, biochemical pathways and their preponderance to contain tandem repeats.

Conclusions

We find significant differences in the evolution between ordered and disordered regions of proteins. Most importantly we find that disorder promoting amino acids are more conserved in IDRs, indicating that in some cases not only amino acid composition but the specific sequence is important for function. This conjecture is also reinforced by the observation that for of our data set IDRs evolve more slowly than the ordered parts of the proteins, while we still support the common view that IDRs in general evolve more quickly. The improvement in model fit indicates a possible improvement for various types of analyses e.g. de novo disorder prediction using a phylogenetic Hidden Markov Model based on our matrices showed a performance similar to other disorder predictors.     

Optimizing long intrinsic disorder predictors with protein evolutionary information

Protein existing as an ensemble of structures, called intrinsically disordered, has been shown to be responsible for a wide variety of biological functions and to be common in nature. Here we focus on improving sequence-based predictions of long (>30 amino acid residues) regions lacking specific 3-D structure by means of four new neural-network-based Predictors Of Natural Disordered Regions (PONDRs): VL3, VL3H, VL3P, and VL3E. PONDR VL3 used several features from a previously introduced PONDR VL2, but benefitted from optimized predictor models and a slightly larger (152 vs. 145) set of disordered proteins that were cleaned of mislabeling errors found in the smaller set. PONDR VL3H utilized homologues of the disordered proteins in the training stage, while PONDR VL3P used attributes derived from sequence profiles obtained by PSI-BLAST searches. The measure of accuracy was the average between accuracies on disordered and ordered protein regions. By this measure, the 30-fold cross-validation accuracies of VL3, VL3H, and VL3P were, respectively, 83.6 +/- 1.4%, 85.3 +/- 1.4%, and 85.2 +/- 1.5%. By combining VL3H and VL3P, the resulting PONDR VL3E achieved an accuracy of 86.7 +/- 1.4%. This is a significant improvement over our previous PONDRs VLXT (71.6 +/- 1.3%) and VL2 (80.9 +/- 1.4%). The new disorder predictors with the corresponding datasets are freely accessible through the web server at http://www.ist.temple.edu/disprot.     

Using Bayesian multinomial classifier to predict whether a given protein sequence is intrinsically disordered

Intrinsically disordered proteins (IDPs) lack a well-defined three-dimensional structure under physiological conditions. Intrinsic disorder is a common phenomenon, particularly in multicellular eukaryotes, and is responsible for important protein functions including regulation and signaling. Many disease-related proteins are likely to be intrinsically disordered or to have disordered regions. In this paper, a new predictor model based on the Bayesian classification methodology is introduced to predict for a given protein or protein region if it is intrinsically disordered or ordered using only its primary sequence. The method allows to incorporate length-dependent amino acid compositional differences of disordered regions by including separate statistical representations for short, middle and long disordered regions. The predictor was trained on the constructed data set of protein regions with known structural properties. In a Jack-knife test, the predictor achieved the sensitivity of 89.2% for disordered and 81.4% for ordered regions. Our method outperformed several reported predictors when evaluated on the previously published data set of Prilusky et al. [2005. FoldIndex: a simple tool to predict whether a given protein sequence is intrinsically unfolded. Bioinformatics 21 (16), 3435-3438]. Further strength of our approach is the ease of implementation.     

Local structural disorder imparts plasticity on linear motifs

MOTIVATION: The dynamic nature of protein interaction networks requires fast and transient molecular switches. The underlying recognition motifs (linear motifs, LMs) are usually short and evolutionarily variable segments, which in several cases, such as phosphorylation sites or SH3-binding regions, fall into locally disordered regions. We probed the generality of this phenomenon by predicting the intrinsic disorder of all LM-containing proteins enlisted in the Eukaryotic Linear Motif (ELM) database. RESULTS: We demonstrated that LMs in average are embedded in locally unstructured regions, while their amino acid composition and charge/hydropathy properties exhibit a mixture characteristic of folded and disordered proteins. Overall, LMs are constructed by grafting a few specificity-determining residues favoring structural order on a highly flexible carrier region. These results establish a connection between LMs and molecular recognition elements of intrinsically unstructured proteins (IUPs), which realize a non-conventional mode of partner binding mostly in regulatory functions.     

Prediction of short linear protein binding regions

Short linear motifs in proteins (typically 3-12 residues in length) play key roles in protein-protein interactions by frequently binding specifically to peptide binding domains within interacting proteins. Their tendency to be found in disordered segments of proteins has meant that they have often been overlooked. Here we present SLiMPred (short linear motif predictor), the first general de novo method designed to computationally predict such regions in protein primary sequences independent of experimentally defined homologs and interactors. The method applies machine learning techniques to predict new motifs based on annotated instances from the Eukaryotic Linear Motif database, as well as structural, biophysical, and biochemical features derived from the protein primary sequence. We have integrated these data sources and benchmarked the predictive accuracy of the method, and found that it performs equivalently to a predictor of protein binding regions in disordered regions, in addition to having predictive power for other classes of motif sites such as polyproline II helix motifs and short linear motifs lying in ordered regions. It will be useful in predicting peptides involved in potential protein associations and will aid in the functional characterization of proteins, especially of proteins lacking experimental information on structures and interactions. We conclude that, despite the diversity of motif sequences and structures, SLiMPred is a valuable tool for prioritizing potential interaction motifs in proteins.     

Length-dependent prediction of protein intrinsic disorder

Background  

Due to the functional importance of intrinsically disordered proteins or protein regions, prediction of intrinsic protein disorder from amino acid sequence has become an area of active research as witnessed in the 6th experiment on Critical Assessment of Techniques for Protein Structure Prediction (CASP6). Since the initial work by Romero et al. (Identifying disordered regions in proteins from amino acid sequences, IEEE Int. Conf. Neural Netw., 1997), our group has developed several predictors optimized for long disordered regions (>30 residues) with prediction accuracy exceeding 85%. However, these predictors are less successful on short disordered regions (≤30 residues). A probable cause is a length-dependent amino acid compositions and sequence properties of disordered regions.     

Flavors of protein disorder

Intrinsically disordered proteins are characterized by long regions lacking 3-D structure in their native states, yet they have been so far associated with 28 distinguishable functions. Previous studies showed that protein predictors trained on disorder from one type of protein often achieve poor accuracy on disorder of proteins of a different type, thus indicating significant differences in sequence properties among disordered proteins. Important biological problems are identifying different types, or flavors, of disorder and examining their relationships with protein function. Innovative use of computational methods is needed in addressing these problems due to relative scarcity of experimental data and background knowledge related to protein disorder. We developed an algorithm that partitions protein disorder into flavors based on competition among increasing numbers of predictors, with prediction accuracy determining both the number of distinct predictors and the partitioning of the individual proteins. Using 145 variously characterized proteins with long (>30 amino acids) disordered regions, 3 flavors, called V, C, and S, were identified by this approach, with the V subset containing 52 segments and 7743 residues, C containing 39 segments and 3402 residues, and S containing 54 segments and 5752 residues. The V, C, and S flavors were distinguishable by amino acid compositions, sequence locations, and biological function. For the sequences in SwissProt and 28 genomes, their protein functions exhibit correlations with the commonness and usage of different disorder flavors, suggesting different flavor-function sets across these protein groups. Overall, the results herein support the flavor-function approach as a useful complement to structural genomics as a means for automatically assigning possible functions to sequences.     

The Ising model for prediction of disordered residues from protein sequence alone

Intrinsically disordered regions serve as molecular recognition elements, which play an important role in the control of many cellular processes and signaling pathways. It is useful to be able to predict positions of disordered residues and disordered regions in protein chains using protein sequence alone. A new method (IsUnstruct) based on the Ising model for prediction of disordered residues from protein sequence alone has been developed. According to this model, each residue can be in one of two states: ordered or disordered. The model is an approximation of the Ising model in which the interaction term between neighbors has been replaced by a penalty for changing between states (the energy of border). The IsUnstruct has been compared with other available methods and found to perform well. The method correctly finds 77% of disordered residues as well as 87% of ordered residues in the CASP8 database, and 72% of disordered residues as well as 85% of ordered residues in the DisProt database.     

PONDR-FIT: A meta-predictor of intrinsically disordered amino acids

Protein intrinsic disorder is becoming increasingly recognized in proteomics research. While lacking structure, many regions of disorder have been associated with biological function. There are many different experimental methods for characterizing intrinsically disordered proteins and regions; nevertheless, the prediction of intrinsic disorder from amino acid sequence remains a useful strategy especially for many large-scale proteomic investigations. Here we introduced a consensus artificial neural network (ANN) prediction method, which was developed by combining the outputs of several individual disorder predictors. By eight-fold cross-validation, this meta-predictor, called PONDR-FIT, was found to improve the prediction accuracy over a range of 3 to 20% with an average of 11% compared to the single predictors, depending on the datasets being used. Analysis of the errors shows that the worst accuracy still occurs for short disordered regions with less than ten residues, as well as for the residues close to order/disorder boundaries. Increased understanding of the underlying mechanism by which such meta-predictors give improved predictions will likely promote the further development of protein disorder predictors. Access to PONDR-FIT is available at www.disprot.org.     

Genome‐scale prediction of proteins with long intrinsically disordered regions

Proteins with long disordered regions (LDRs), defined as having 30 or more consecutive disordered residues, are abundant in eukaryotes, and these regions are recognized as a distinct class of biologically functional domains. LDRs facilitate various cellular functions and are important for target selection in structural genomics. Motivated by the lack of methods that directly predict proteins with LDRs, we designed Super‐fast predictor of proteins with Long Intrinsically DisordERed regions (SLIDER). SLIDER utilizes logistic regression that takes an empirically chosen set of numerical features, which consider selected physicochemical properties of amino acids, sequence complexity, and amino acid composition, as its inputs. Empirical tests show that SLIDER offers competitive predictive performance combined with low computational cost. It outperforms, by at least a modest margin, a comprehensive set of modern disorder predictors (that can indirectly predict LDRs) and is 16 times faster compared to the best currently available disorder predictor. Utilizing our time‐efficient predictor, we characterized abundance and functional roles of proteins with LDRs over 110 eukaryotic proteomes. Similar to related studies, we found that eukaryotes have many (on average 30.3%) proteins with LDRs with majority of proteomes having between 25 and 40%, where higher abundance is characteristic to proteomes that have larger proteins. Our first‐of‐its‐kind large‐scale functional analysis shows that these proteins are enriched in a number of cellular functions and processes including certain binding events, regulation of catalytic activities, cellular component organization, biogenesis, biological regulation, and some metabolic and developmental processes. A webserver that implements SLIDER is available at http://biomine.ece.ualberta.ca/SLIDER/ .Proteins 2014; 82:145–158. © 2013 Wiley Periodicals, Inc.