Prediction of protein structural class using tri-gram probabilities of position-specific scoring matrix and recursive feature elimination

Recent Advances in the Prediction of Protein Structural Classes: Feature Descriptors and Machine Learning Algorithms

In the postgenomic age, rapid growth in the number of sequence-known proteins has been accompanied by much slower growth in the number of structure-known proteins (as a result of experimental limitations), and a widening gap between the two is evident. Because protein function is linked to protein structure, successful prediction of protein structure is of significant importance in protein function identification. Foreknowledge of protein structural class can help improve protein structure prediction with significant medical and pharmaceutical implications. Thus, a fast, suitable, reliable, and reasonable computational method for protein structural class prediction has become pivotal in bioinformatics. Here, we review recent efforts in protein structural class prediction from protein sequence, with particular attention paid to new feature descriptors, which extract information from protein sequence, and the use of machine learning algorithms in both feature selection and the construction of new classification models. These new feature descriptors include amino acid composition, sequence order, physicochemical properties, multiprofile Bayes, and secondary structure-based features. Machine learning methods, such as artificial neural networks (ANNs), support vector machine (SVM), K-nearest neighbor (KNN), random forest, deep learning, and examples of their application are discussed in detail. We also present our view on possible future directions, challenges, and opportunities for the applications of machine learning algorithms for prediction of protein structural classes.

Recursive Feature Elimination by Sensitivity Testing

There is great interest in methods to improve human insight into trained non-linear models. Leading approaches include producing a ranking of the most relevant features, a non-trivial task for non-linear models. We show theoretically and empirically the benefit of a novel version of recursive feature elimination (RFE) as often used with SVMs; the key idea is a simple twist on the kinds of sensitivity testing employed in computational learning theory with membership queries (e.g., [1]). With membership queries, one can check whether changing the value of a feature in an example changes the label. In the real-world, we usually cannot get answers to such queries, so our approach instead makes these queries to a trained (imperfect) non-linear model. Because SVMs are widely used in bioinformatics, our empirical results use a real-world cancer genomics problem; because ground truth is not known for this task, we discuss the potential insights provided. We also evaluate on synthetic data where ground truth is known.

ProTstab - predictor for cellular protein stability

BACKGROUND

Stability is one of the most fundamental intrinsic characteristics of proteins and can be determined with various methods. Characterization of protein properties does not keep pace with increase in new sequence data and therefore even basic properties are not known for far majority of identified proteins. There have been some attempts to develop predictors for protein stabilities; however, they have suffered from small numbers of known examples.

RESULTS

We took benefit of results from a recently developed cellular stability method, which is based on limited proteolysis and mass spectrometry, and developed a machine learning method using gradient boosting of regression trees. ProTstab method has high performance and is well suited for large scale prediction of protein stabilities.

CONCLUSIONS

The Pearson's correlation coefficient was 0.793 in 10-fold cross validation and 0.763 in independent blind test. The corresponding values for mean absolute error are 0.024 and 0.036, respectively. Comparison with a previously published method indicated ProTstab to have superior performance. We used the method to predict stabilities of all the remaining proteins in the entire human proteome and then correlated the predicted stabilities to protein chain lengths of isoforms and to localizations of proteins.

iAPSL-IF: Identification of Apoptosis Protein Subcellular Location Using Integrative Features Captured from Amino Acid Sequences

Apoptosis proteins (APs) control normal tissue homeostasis by regulating the balance between cell proliferation and death. The function of APs is strongly related to their subcellular location. To date, computational methods have been reported that reliably identify the subcellular location of APs, however, there is still room for improvement of the prediction accuracy. In this study, we developed a novel method named iAPSL-IF (identification of apoptosis protein subcellular location—integrative features), which is based on integrative features captured from Markov chains, physicochemical property matrices, and position-specific score matrices (PSSMs) of amino acid sequences. The matrices with different lengths were transformed into fixed-length feature vectors using an auto cross-covariance (ACC) method. An optimal subset of the features was chosen using a recursive feature elimination (RFE) algorithm method, and the sequences with these features were trained by a support vector machine (SVM) classifier. Based on three datasets ZD98, CL317, and ZW225, the iAPSL-IF was examined using a jackknife cross-validation test. The resulting data showed that the iAPSL-IF outperformed the known predictors reported in the literature: its overall accuracy on the three datasets was 98.98% (ZD98), 94.95% (CL317), and 97.33% (ZW225), respectively; the Matthews correlation coefficient, sensitivity, and specificity for several classes of subcellular location proteins (e.g., membrane proteins, cytoplasmic proteins, endoplasmic reticulum proteins, nuclear proteins, and secreted proteins) in the datasets were 0.92–1.0, 94.23–100%, and 97.07–100%, respectively. Overall, the results of this study provide a high throughput and sequence-based method for better identification of the subcellular location of APs, and facilitates further understanding of programmed cell death in organisms.

Subcellular localization prediction of apoptosis proteins based on evolutionary information and support vector machine

OBJECTIVES

In this paper, a high-quality sequence encoding scheme is proposed for predicting subcellular location of apoptosis proteins.

METHODS

In the proposed methodology, the novel evolutionary-conservative information is introduced to represent protein sequences. Meanwhile, based on the proportion of golden section in mathematics, position-specific scoring matrix (PSSM) is divided into several blocks. Then, these features are predicted by support vector machine (SVM) and the predictive capability of proposed method is implemented by jackknife test RESULTS: The results show that the golden section method is better than no segmentation method. The overall accuracy for ZD98 and CL317 is 98.98% and 91.11%, respectively, which indicates that our method can play a complimentary role to the existing methods in the relevant areas.

CONCLUSIONS

The proposed feature representation is powerful and the prediction accuracy will be improved greatly, which denotes our method provides the state-of-the-art performance for predicting subcellular location of apoptosis proteins.

Predicting Presynaptic and Postsynaptic Neurotoxins by Developing Feature Selection Technique

Presynaptic and postsynaptic neurotoxins are proteins which act at the presynaptic and postsynaptic membrane. Correctly predicting presynaptic and postsynaptic neurotoxins will provide important clues for drug-target discovery and drug design. In this study, we developed a theoretical method to discriminate presynaptic neurotoxins from postsynaptic neurotoxins. A strict and objective benchmark dataset was constructed to train and test our proposed model. The dipeptide composition was used to formulate neurotoxin samples. The analysis of variance (ANOVA) was proposed to find out the optimal feature set which can produce the maximum accuracy. In the jackknife cross-validation test, the overall accuracy of 94.9% was achieved. We believe that the proposed model will provide important information to study neurotoxins.

Classify vertebrate hemoglobin proteins by incorporating the evolutionary information into the general PseAAC with the hybrid approach

Hemoglobin is an oxygen-binding protein widely present in all kingdoms of life from prokaryotic to eukaryotic, but well established in the vertebrate system. An attempt was made to determine the Vertebrate hemoglobin (VerHb) protein on their animal classifications, based on general pseudo amino acid composition (PseAAC)'s evolutionary profiles and hybrid approach. The support vector machine (SVM) has been applied to develop all models, the prediction results further compared according to their animal classification. The performance of the approaches estimated using five-fold cross-validation techniques. The prediction performance was further investigated by receiver operating characteristic (ROC) and prediction score graphs. The prediction accuracy (ACC), sensitivity (SN) and specificity (SP) were examined to find the accurate predictions on the threshold level. Based on the approach, a web-tool has been developed for identifying the VerHb proteins.

A Novel Feature Extraction Method with Feature Selection to Identify Golgi-Resident Protein Types from Imbalanced Data

The Golgi Apparatus (GA) is a major collection and dispatch station for numerous proteins destined for secretion, plasma membranes and lysosomes. The dysfunction of GA proteins can result in neurodegenerative diseases. Therefore, accurate identification of protein subGolgi localizations may assist in drug development and understanding the mechanisms of the GA involved in various cellular processes. In this paper, a new computational method is proposed for identifying cis-Golgi proteins from trans-Golgi proteins. Based on the concept of Common Spatial Patterns (CSP), a novel feature extraction technique is developed to extract evolutionary information from protein sequences. To deal with the imbalanced benchmark dataset, the Synthetic Minority Over-sampling Technique (SMOTE) is adopted. A feature selection method called Random Forest-Recursive Feature Elimination (RF-RFE) is employed to search the optimal features from the CSP based features and g-gap dipeptide composition. Based on the optimal features, a Random Forest (RF) module is used to distinguish cis-Golgi proteins from trans-Golgi proteins. Through the jackknife cross-validation, the proposed method achieves a promising performance with a sensitivity of 0.889, a specificity of 0.880, an accuracy of 0.885, and a Matthew's Correlation Coefficient (MCC) of 0.765, which remarkably outperforms previous methods. Moreover, when tested on a common independent dataset, our method also achieves a significantly improved performance. These results highlight the promising performance of the proposed method to identify Golgi-resident protein types. Furthermore, the CSP based feature extraction method may provide guidelines for protein function predictions.