Exploiting sequence dependencies in the prediction of peroxisomal proteins

Prediction of peroxisomal matrix proteins generally depends on the presence of one of two distinct motifs at the end of the amino acid sequence. PTS1 peroxisomal proteins have a well conserved tripeptide at the C-terminal end. However, the preceding residues in the sequence arguably play a crucial role in targeting the protein to the peroxisome. Previous work in applying machine learning to the prediction of peroxisomal matrix proteins has failed W capitalize on the full extent of these dependencies. We benchmark a range of machine learning algorithms, and show that a classifier - based on the Support Vector Machine - produces more accurate results when dependencies between the conserved motif and the preceding section are exploited. We publish an updated and rigorously curated data set that results in increased prediction accuracy of most tested models.

Predicting residue-wise contact orders in proteins by support vector regression

Background: The residue-wise contact order (RWCO) describes the sequence separations between the residues of interest and its contacting residues in a protein sequence. It is a new kind of one-dimensional protein structure that represents the extent of long-range contacts and is considered as a generalization of contact order. Together with secondary structure, accessible surface area, the B factor, and contact number, RWCO provides comprehensive and indispensable important information to reconstructing the protein three-dimensional structure from a set of one-dimensional structural properties. Accurately predicting RWCO values could have many important applications in protein three-dimensional structure prediction and protein folding rate prediction, and give deep insights into protein sequence-structure relationships. Results: We developed a novel approach to predict residue-wise contact order values in proteins based on support vector regression (SVR), starting from primary amino acid sequences. We explored seven different sequence encoding schemes to examine their effects on the prediction performance, including local sequence in the form of PSI-BLAST profiles, local sequence plus amino acid composition, local sequence plus molecular weight, local sequence plus secondary structure predicted by PSIPRED, local sequence plus molecular weight and amino acid composition, local sequence plus molecular weight and predicted secondary structure, and local sequence plus molecular weight, amino acid composition and predicted secondary structure. When using local sequences with multiple sequence alignments in the form of PSI-BLAST profiles, we could predict the RWCO distribution with a Pearson correlation coefficient (CC) between the predicted and observed RWCO values of 0.55, and root mean square error (RMSE) of 0.82, based on a well-defined dataset with 680 protein sequences. Moreover, by incorporating global features such as molecular weight and amino acid composition we could further improve the prediction performance with the CC to 0.57 and an RMSE of 0.79. In addition, combining the predicted secondary structure by PSIPRED was found to significantly improve the prediction performance and could yield the best prediction accuracy with a CC of 0.60 and RMSE of 0.78, which provided at least comparable performance compared with the other existing methods. Conclusion: The SVR method shows a prediction performance competitive with or at least comparable to the previously developed linear regression-based methods for predicting RWCO values. In contrast to support vector classification (SVC), SVR is very good at estimating the raw value profiles of the samples. The successful application of the SVR approach in this study reinforces the fact that support vector regression is a powerful tool in extracting the protein sequence-structure relationship and in estimating the protein structural profiles from amino acid sequences.

Structural basis and prediction of substrate specificity in protein serine/threonine kinases

The large number of protein kinases makes it impractical to determine their specificities and substrates experimentally. Using the available crystal structures, molecular modeling, and sequence analyses of kinases and substrates, we developed a set of rules governing the binding of a heptapeptide substrate motif (surrounding the phosphorylation site) to the kinase and implemented these rules in a web-interfaced program for automated prediction of optimal substrate peptides, taking only the amino acid sequence of a protein kinase as input. We show the utility of the method by analyzing yeast cell cycle control and DNA damage checkpoint pathways. Our method is the only available predictive method generally applicable for identifying possible substrate proteins for protein serine/threonine kinases and helps in silico construction of signaling pathways. The accuracy of prediction is comparable to the accuracy of data from systematic large-scale experimental approaches.

Reducing the time period of steady state does not affect the accuracy of energy expenditure measurements by indirect calorimetry

Achievement of steady state during indirect calorimetry measurements of resting energy expenditure (REE) is necessary to reduce error and ensure accuracy in the measurement. Steady state is often defined as 5 consecutive min (5-min SS) during which oxygen consumption and carbon dioxide production vary by +/-10%. These criteria, however, are stringent and often difficult to satisfy. This study aimed to assess whether reducing the time period for steady state (4-min SS or 3-min SS) produced measurements of REE that were significantly different from 5-min SS. REE was measured with the use of open-circuit indirect calorimetry in 39 subjects, of whom only 21 (54%) met the 5-min SS criteria. In these 21 subjects, median biases in REE between 5-min SS and 4-min SS and between 5-min SS and 3-min SS were 0.1 and 0.01%, respectively. For individuals, 4-min SS measured REE within a clinically acceptable range of +/-2% of 5-min SS, whereas 3-min SS measured REE within a range of -2-3% of 5-min SS. Harris-Benedict prediction equations estimated REE for individuals within +/-20-30% of 5-min SS. Reducing the time period of steady state to 4 min produced measurements of REE for individuals that were within clinically acceptable, predetermined limits. The limits of agreement for 3-min SS fell outside the predefined limits of +/-2%; however, both 4-min SS and 3-min SS criteria greatly increased the proportion of subjects who satisfied steady state within smaller limits than would be achieved if relying on prediction equations.

Protein contact prediction using patterns of correlation

We describe a new method for using neural networks to predict residue contact pairs in a protein. The main inputs to the neural network are a set of 25 measures of correlated mutation between all pairs of residues in two windows of size 5 centered on the residues of interest. While the individual pair-wise correlations are a relatively weak predictor of contact, by training the network on windows of correlation the accuracy of prediction is significantly improved. The neural network is trained on a set of 100 proteins and then tested on a disjoint set of 1033 proteins of known structure. An average predictive accuracy of 21.7% is obtained taking the best L/2 predictions for each protein, where L is the sequence length. Taking the best L/10 predictions gives an average accuracy of 30.7%. The predictor is also tested on a set of 59 proteins from the CASP5 experiment. The accuracy is found to be relatively consistent across different sequence lengths, but to vary widely according to the secondary structure. Predictive accuracy is also found to improve by using multiple sequence alignments containing many sequences to calculate the correlations. (C) 2004 Wiley-Liss, Inc.

Prediction of high-pressure adsorption equilibrium of supercritical gases using density functional theory

In this paper, we present the results of the prediction of the high-pressure adsorption equilibrium of supercritical. gases (Ar, N-2, CH4, and CO2) on various activated carbons (BPL, PCB, and Norit R1 extra) at various temperatures using a density-functional-theory-based finite wall thickness (FWT) model. Pore size distribution results of the carbons are taken from our recent previous work 1,2 using this approach for characterization. To validate the model, isotherms calculated from the density functional theory (DFT) approach are comprehensively verified against those determined by grand canonical Monte Carlo (GCMC) simulation, before the theoretical adsorption isotherms of these investigated carbons calculated by the model are compared with the experimental adsorption measurements of the carbons. We illustrate the accuracy and consistency of the FWT model for the prediction of adsorption isotherms of the all investigated gases. The pore network connectivity problem occurring in the examined carbons is also discussed, and on the basis of the success of the predictions assuming a similar pore size distribution for accessible and inaccessible regions, it is suggested that this is largely related to the disordered nature of the carbon.

Relative accuracy and predictive ability of direct valuation methods, price to aggregate earnings method and a hybrid approach

In this paper, we assess the relative performance of the direct valuation method and industry multiplier models using 41 435 firm-quarter Value Line observations over an 11 year (1990–2000) period. Results from both pricingerror and return-prediction analyses indicate that direct valuation yields lower percentage pricing errors and greater return prediction ability than the forward price to aggregated forecasted earnings multiplier model. However, a simple hybrid combination of these two methods leads to more accurate intrinsic value estimates, compared to either method used in isolation. It would appear that fundamental analysis could benefit from using one approach as a check on the other.

Prediction of subcellular localization using sequence-biased recurrent networks

Motivation: Targeting peptides direct nascent proteins to their specific subcellular compartment. Knowledge of targeting signals enables informed drug design and reliable annotation of gene products. However, due to the low similarity of such sequences and the dynamical nature of the sorting process, the computational prediction of subcellular localization of proteins is challenging. Results: We contrast the use of feed forward models as employed by the popular TargetP/SignalP predictors with a sequence-biased recurrent network model. The models are evaluated in terms of performance at the residue level and at the sequence level, and demonstrate that recurrent networks improve the overall prediction performance. Compared to the original results reported for TargetP, an ensemble of the tested models increases the accuracy by 6 and 5% on non-plant and plant data, respectively.

Prediction of protein continuum secondary structure with probabilistic models based on NMR solved structures

Background: The structure of proteins may change as a result of the inherent flexibility of some protein regions. We develop and explore probabilistic machine learning methods for predicting a continuum secondary structure, i.e. assigning probabilities to the conformational states of a residue. We train our methods using data derived from high-quality NMR models. Results: Several probabilistic models not only successfully estimate the continuum secondary structure, but also provide a categorical output on par with models directly trained on categorical data. Importantly, models trained on the continuum secondary structure are also better than their categorical counterparts at identifying the conformational state for structurally ambivalent residues. Conclusion: Cascaded probabilistic neural networks trained on the continuum secondary structure exhibit better accuracy in structurally ambivalent regions of proteins, while sustaining an overall classification accuracy on par with standard, categorical prediction methods.

Prediction of HLA-DQ3.2 beta ligands: evidence of multiple registers in class II binding peptides

Motivation: While processing of MHC class II antigens for presentation to helper T-cells is essential for normal immune response, it is also implicated in the pathogenesis of autoimmune disorders and hypersensitivity reactions. Sequence-based computational techniques for predicting HLA-DQ binding peptides have encountered limited success, with few prediction techniques developed using three-dimensional models. Methods: We describe a structure-based prediction model for modeling peptide-DQ3.2 beta complexes. We have developed a rapid and accurate protocol for docking candidate peptides into the DQ3.2 beta receptor and a scoring function to discriminate binders from the background. The scoring function was rigorously trained, tested and validated using experimentally verified DQ3.2 beta binding and non-binding peptides obtained from biochemical and functional studies. Results: Our model predicts DQ3.2 beta binding peptides with high accuracy [area under the receiver operating characteristic (ROC) curve A(ROC) > 0.90], compared with experimental data. We investigated the binding patterns of DQ3.2 beta peptides and illustrate that several registers exist within a candidate binding peptide. Further analysis reveals that peptides with multiple registers occur predominantly for high-affinity binders.

Evaluation and comparison of mammalian subcellular localization prediction methods

Background: Determination of the subcellular location of a protein is essential to understanding its biochemical function. This information can provide insight into the function of hypothetical or novel proteins. These data are difficult to obtain experimentally but have become especially important since many whole genome sequencing projects have been finished and many resulting protein sequences are still lacking detailed functional information. In order to address this paucity of data, many computational prediction methods have been developed. However, these methods have varying levels of accuracy and perform differently based on the sequences that are presented to the underlying algorithm. It is therefore useful to compare these methods and monitor their performance. Results: In order to perform a comprehensive survey of prediction methods, we selected only methods that accepted large batches of protein sequences, were publicly available, and were able to predict localization to at least nine of the major subcellular locations (nucleus, cytosol, mitochondrion, extracellular region, plasma membrane, Golgi apparatus, endoplasmic reticulum (ER), peroxisome, and lysosome). The selected methods were CELLO, MultiLoc, Proteome Analyst, pTarget and WoLF PSORT. These methods were evaluated using 3763 mouse proteins from SwissProt that represent the source of the training sets used in development of the individual methods. In addition, an independent evaluation set of 2145 mouse proteins from LOCATE with a bias towards the subcellular localization underrepresented in SwissProt was used. The sensitivity and specificity were calculated for each method and compared to a theoretical value based on what might be observed by random chance. Conclusion: No individual method had a sufficient level of sensitivity across both evaluation sets that would enable reliable application to hypothetical proteins. All methods showed lower performance on the LOCATE dataset and variable performance on individual subcellular localizations was observed. Proteins localized to the secretory pathway were the most difficult to predict, while nuclear and extracellular proteins were predicted with the highest sensitivity.

Customer information system data pre-processing with feature selection techniques for non-technical losses prediction in an electricity market

Non-technical losses (NTL) identification and prediction are important tasks for many utilities. Data from customer information system (CIS) can be used for NTL analysis. However, in order to accurately and efficiently perform NTL analysis, the original data from CIS need to be pre-processed before any detailed NTL analysis can be carried out. In this paper, we propose a feature selection based method for CIS data pre-processing in order to extract the most relevant information for further analysis such as clustering and classifications. By removing irrelevant and redundant features, feature selection is an essential step in data mining process in finding optimal subset of features to improve the quality of result by giving faster time processing, higher accuracy and simpler results with fewer features. Detailed feature selection analysis is presented in the paper. Both time-domain and load shape data are compared based on the accuracy, consistency and statistical dependencies between features.