Amino Acid Composition Features Research Articles

A Sequence-Based Predictor of Zika Virus Proteins Developed by Integration of PseAAC and Statistical Moments

ZIKV has been a well-known global threat, which hits almost all of the American countries and posed a serious threat to the entire globe in 2016. The first outbreak of ZIKV was reported in 2007 in the Pacific area, followed by another severe outbreak, which occurred in 2013/2014 and subsequently, ZIKV spread to all other Pacific islands. A broad spectrum of ZIKV associated neurological malformations in neonates and adults has driven this deadly virus into the limelight. Though tremendous efforts have been focused on understanding the molecular basis of ZIKV, the viral proteins of ZIKV have still not been studied extensively. Herein, we report the first and the novel predictor for the identification of ZIKV proteins. We have employed Chou's pseudo amino acid composition (PseAAC), statistical moments and various position-based features. The predictor is validated through 10-fold cross-validation and Jackknife testing. In 10- fold cross-validation, 94.09% accuracy, 93.48% specificity, 94.20% sensitivity and 0.80 MCC were achieved while in Jackknife testing, 96.62% accuracy, 94.57% specificity, 97.00% sensitivity and 0.88 MCC were achieved. Thus, ZIKVPred-PseAAC can help in predicting the ZIKV proteins efficiently and accurately and can provide baseline data for the discovery of new drugs and biomarkers against ZIKV.

HMMPred: Accurate Prediction of DNA-Binding Proteins Based on HMM Profiles and XGBoost Feature Selection.

Prediction of DNA-binding proteins (DBPs) has become a popular research topic in protein science due to its crucial role in all aspects of biological activities. Even though considerable efforts have been devoted to developing powerful computational methods to solve this problem, it is still a challenging task in the field of bioinformatics. A hidden Markov model (HMM) profile has been proved to provide important clues for improving the prediction performance of DBPs. In this paper, we propose a method, called HMMPred, which extracts the features of amino acid composition and auto- and cross-covariance transformation from the HMM profiles, to help train a machine learning model for identification of DBPs. Then, a feature selection technique is performed based on the extreme gradient boosting (XGBoost) algorithm. Finally, the selected optimal features are fed into a support vector machine (SVM) classifier to predict DBPs. The experimental results tested on two benchmark datasets show that the proposed method is superior to most of the existing methods and could serve as an alternative tool to identify DBPs.

SulSite-GTB: identification of protein S-sulfenylation sites by fusing multiple feature information and gradient tree boosting

Protein cysteine S-sulfenylation is an essential and reversible post-translational modification that plays a crucial role in transcriptional regulation, stress response, cell signaling and protein function. Studies have shown that S-sulfenylation is involved in many human diseases such as cancer, diabetes and arteriosclerosis. However, experimental identification of protein S-sulfenylation sites is generally expensive and time-consuming. In this study, we proposed a new protein S-sulfenylation sites prediction method SulSite-GTB. First, fusion of amino acid composition, dipeptide composition, encoding based on grouped weight, K nearest neighbors, position-specific amino acid propensity, position-weighted amino acid composition and pseudo-position specific score matrix feature extraction to obtain the initial feature space. Secondly, we use the synthetic minority oversampling technique (SMOTE) algorithm to process the class imbalance data, and the least absolute shrinkage and selection operator (LASSO) are employed to remove the redundant and irrelevant features. Finally, the optimal feature subset is input into the gradient tree boosting classifier to predict the S-sulfenylation sites, and the five-fold cross-validation and independent test set method are used to evaluate the prediction performance of the model. Experimental results showed the overall prediction accuracy is 92.86% and 88.53%, respectively, and the AUC values are 0.9706 and 0.9425, respectively, on the training set and the independent test set. Compared with other prediction methods, the results show that the proposed method SulSite-GTB is significantly superior to other state-of-the-art methods and provides a new idea for the prediction of post-translational modification sites of other proteins. The source code and all datasets are available at https://github.com/QUST-AIBBDRC/SulSite-GTB/ .

Comparative Analysis of Thermophilic Alpha-Amylase Using In Silico Approach

The organism that can live outside of the ‘normal’ range of at least one environmental factor is an extremophile. A thermophile is an organism a type of extremophile that thrives at relatively high temperatures, between 45 and 122 °C. There are several mechanisms used thermophiles hyperthermophiles to maintain the stability of the proteins for their survival. To study this stability there are many Bioinformatics tools available. The amino acid composition is believed to alter the intrinsic stability of proteins. Several comparative studies of the primary sequence obtained from 97 thermophilic organisms have been carried out between Mesophilic and Thermophilic Alpha-amylases to understand the amino acid composition for the thermostability of proteins. This work presents some generalized features of amino acid composition found in thermophilic proteins, including an increase in residue hydrophobicity, a decrease in uncharged polar residues, an increase in charged residues, an increase in aromatic residues. The differences of amino acids composition between thermophilic and mesophilic proteins are related to some properties of amino acids. These features provide guidelines for engineering mesophilic protein to thermophilic protein.

SMOPredT4SE: An Effective Prediction of Bacterial Type IV Secreted Effectors Using SVM Training With SMO

Various bacterial pathogens can deliver their secreted effectors to host cells via type IV secretion system (T4SS) and cause host diseases. Since T4SS secreted effectors (T4SEs) play important roles in the interaction between pathogens and host, identifying T4SEs is crucial to understanding of the pathogenic mechanism of T4SS. We established an effective predictor called SMOPredT4SE to identify T4SEs from protein sequences. SMOPredT4SE employed combination features of series correlation pseudo amino acid composition and position-specific scoring matrix to present protein sequences, and employed support vector machines (SVM) training with sequential minimal optimization (SMO) arithmetic to train the prediction model (To distinguish it from the traditional SVM, we will abbreviate it as SMO later). In the 5-fold cross-validation test, SMOPredT4SE's overall accuracy was 95.6%. Experiments on comparison with other feature, classifiers, and existing methods are conducted. Experimental results show the effectiveness of SMOPredT4SE in predicting T4SEs.

Comparative genomics of hepatitis A virus, hepatitis C virus, and hepatitis E virus provides insights into the evolutionary history of Hepatovirus species.

The intraspecies genomic diversity of the single‐strand RNA (+) virus species hepatitis A virus (Hepatovirus), hepatitis C virus (Hepacivirus), and hepatitis E virus (Orthohepevirus) was compared. These viral species all can cause liver inflammation (hepatitis), but share no gene similarity. The codon usage of human hepatitis A virus (HAV) is suboptimal for replication in its host, a characteristic it shares with taxonomically related rodent, simian, and bat hepatitis A virus species. We found this codon usage to be strikingly similar to that of Triatoma virus that infects blood‐sucking kissing bugs. The codon usage of that virus is well adapted to its insect host. The codon usage of HAV is also similar to other invertebrate viruses of various taxonomic families. An evolutionary ancestor of HAV and related virus species is hypothesized to be an insect virus that underwent a host jump to infect mammals. The similarity between HAV and invertebrate viruses goes beyond codon usage, as they also share amino acid composition characteristics, while not sharing direct sequence homology. In contrast, hepatitis C virus and hepatitis E virus are highly similar in codon usage preference, nucleotide composition, and amino acid composition, and share these characteristics with Human pegivirus A, West Nile virus, and Zika virus. We present evidence that these observations are only partly explained by differences in nucleotide composition of the complete viral codon regions. We consider the combination of nucleotide composition, amino acid composition, and codon usage preference suitable to provide information on possible evolutionary similarities between distant virus species that cannot be investigated by phylogeny.

Plant-mSubP: a computational framework for the prediction of single- and multi-target protein subcellular localization using integrated machine-learning approaches.

The subcellular localization of proteins is very important for characterizing its function in a cell. Accurate prediction of the subcellular locations in computational paradigm has been an active area of interest. Most of the work has been focused on single localization prediction. Only few studies have discussed the multi-target localization, but have not achieved good accuracy so far; in plant sciences, very limited work has been done. Here we report the development of a novel tool Plant-mSubP, which is based on integrated machine learning approaches to efficiently predict the subcellular localizations in plant proteomes. The proposed approach predicts with high accuracy 11 single localizations and three dual locations of plant cell. Several hybrid features based on composition and physicochemical properties of a protein such as amino acid composition, pseudo amino acid composition, auto-correlation descriptors, quasi-sequence-order descriptors and hybrid features are used to represent the protein. The performance of the proposed method has been assessed through a training set as well as an independent test set. Using the hybrid feature of the pseudo amino acid composition, N-Center-C terminal amino acid composition and the dipeptide composition (PseAAC-NCC-DIPEP), an overall accuracy of 81.97 %, 84.75 % and 87.88 % is achieved on the training data set of proteins containing the single-label, single- and dual-label combined, and dual-label proteins, respectively. When tested on the independent data, an accuracy of 64.36 %, 64.84 % and 81.08 % is achieved on the single-label, single- and dual-label, and dual-label proteins, respectively. The prediction models have been implemented on a web server available at http://bioinfo.usu.edu/Plant-mSubP/. The results indicate that the proposed approach is comparable to the existing methods in single localization prediction and outperforms all other existing tools when compared for dual-label proteins. The prediction tool will be a useful resource for better annotation of various plant proteomes.

Ragp: Pipeline for mining of plant hydroxyproline-rich glycoproteins with implementation in R.

Hydroxyproline-rich glycoproteins (HRGPs) are one of the most complex families of macromolecules found in plants, due to the diversity of glycans decorating the protein backbone, as well as the heterogeneity of the protein backbones. While this diversity is responsible for a wide array of physiological functions associated with HRGPs, it hinders attempts for homology based identification. Current approaches, based on identifying sequences with characteristic motifs and biased amino acid composition, are limited to prototypical sequences. Ragp is an R package for mining and analysis of HRGPs, with emphasis on arabinogalactan proteins. The ragp filtering pipeline exploits one of the HRGPs key features, the presence of hydroxyprolines which represent glycosylation sites. Main package features include prediction of proline hydroxylation sites, amino acid motif and bias analyses, efficient communication with web servers for prediction of N-terminal signal peptides, glycosylphosphatidylinositol modification sites and disordered regions and the ability to annotate sequences through hmmscan and subsequent GO enrichment, based on predicted Pfam domains. As such, ragp extends R's rich ecosystem for high-throughput sequence data analyses. The ragp R package is available under the MIT Open Source license and is freely available to download from GitHub at: https://github.com/missuse/ragp.

A Random Forest Sub-Golgi Protein Classifier Optimized via Dipeptide and Amino Acid Composition Features.

To gain insight into the malfunction of the Golgi apparatus and its relationship to various genetic and neurodegenerative diseases, the identification of sub-Golgi proteins, both cis-Golgi and trans-Golgi proteins, is of great significance. In this study, a state-of-art random forests sub-Golgi protein classifier, rfGPT, was developed. The rfGPT used 2-gap dipeptide and split amino acid composition for the feature vectors and was combined with the synthetic minority over-sampling technique (SMOTE) and an analysis of variance (ANOVA) feature selection method. The rfGPT was trained on a sub-Golgi protein sequence data set (137 sequences), with sequence identity less than 25%. For the optimal rfGPT classifier with 93 features, the accuracy (ACC) was 90.5%; the Matthews correlation coefficient (MCC) was 0.811; the sensitivity (Sn) was 92.6%; and the specificity (Sp) was 88.4%. The independent testing scores for the rfGPT were ACC = 90.6%; MCC = 0.696; Sn = 96.1%; and Sp = 69.2%. Although the independent testing accuracy was 4.4% lower than that for the best reported sub-Golgi classifier trained on a data set with 40% sequence identity (304 sequences), the rfGPT is currently the top sub-Golgi protein predictor utilizing feature vectors without any position-specific scoring matrix and its derivative features. Therefore, the rfGPT is a more practical tool, because no sequence alignment is required with tens of millions of protein sequences. To date, the rfGPT is the Golgi classifier with the best independent testing scores, optimized by training on smaller benchmark data sets. Feature importance analysis proves that the non-polar and aliphatic residues composition, the (aromatic residues) + (non-polar, aliphatic residues) dipeptide and aromatic residues composition between NH2-termial and COOH-terminal of protein sequences are the three top biological features for distinguishing the sub-Golgi proteins.

Improved Prediction of Cell-Penetrating Peptides via Effective Orchestrating Amino Acid Composition Feature Representation

Cell-penetrating peptides (CPPs) promote the transport of pharmacologically active molecules, such as nanoparticles, plasmid DNA and short interfering RNA. Accurate prediction of new CPPs is a prerequisite for in-depth study of such molecules. Biological experimental predictions can provide an accurate description of the penetrating properties of CPPs. However, predicting CPPs by wet laboratory experiments is both resource-intensive and time-consuming. Therefore, the development of effective calculation method prediction has become an important topic in the study of CPPs. Recently, numerous methods developed for predicting CPPs use amino acid composition, alone and the accuracies of such methods have been limited. In this study, we proposed a new CPP prediction framework, which integrates four amino acid composition features, and utilizes these features to help train Support Vector Machine (SVM) model as a classifier to predict CPPs. When performing on the training dataset CPP924, the proposed method achieves an accuracy of 92.3%, which is significantly better than the state-of-the-art methods. These results suggest that the framework can orchestrate various amino acid composition features predicted models flexibly with good performances.

A computational method for prediction of xylanase enzymes activity in strains of Bacillus subtilis based on pseudo amino acid composition features.

Xylanases are hydrolytic enzymes which based on physicochemical properties, structure, mode of action and substrate specificities are classified into various glycoside hydrolase (GH) families. The purpose of this study is to show that the activity of the members of the xylanase family in the specified pH and temperature conditions can be computationally predicted. The proposed computational regression model was trained and tested with the Pseudo Amino Acid Composition (PseAAC) features extracted solely from the amino acid sequences of enzymes. The xylanases with experimentally determined activities were used as the training dataset to adjust the model parameters. To develop the model, 41 strains of Bacillus subtilis isolated from field soil were screened. From them, 28 strains with the highest halo diameter were selected for further studies. The performance of the model for prediction of xylanase activity was evaluated in three different temperature and pH conditions using stratified cross-validation and jackknife methods. The trained model can be used for determining the activity of newly found xylanases in the specified condition. Such computational models help to scale down the experimental costs and save time by identifying enzymes with appropriate activity for scientific and industrial usage. Our methodology for activity prediction of xylanase enzymes can be potentially applied to the members of the other enzyme families. The availability of sufficient experimental data in specified pH and temperature conditions is a prerequisite for training the learning model and to achieve high accuracy.

Quantification of Intrinsically Disordered Proteins: A Problem Not Fully Appreciated.

Protein quantification is essential in a great variety of biochemical assays, yet the inherent systematic errors associated with the concentration determination of intrinsically disordered proteins (IDPs) using classical methods are hardly appreciated. Routinely used assays for protein quantification, such as the Bradford assay or ultraviolet absorbance at 280 nm, usually seriously misestimate the concentrations of IDPs due to their distinct and variable amino acid composition. Therefore, dependable method(s) have to be worked out/adopted for this task. By comparison to elemental analysis as the gold standard, we show through the example of four globular proteins and nine IDPs that the ninhydrin assay and the commercial QubitTM Protein Assay provide reliable data on IDP quantity. However, as IDPs can show extreme variation in amino acid composition and physical features not necessarily covered by our examples, even these techniques should only be used for IDPs following standardization. The far-reaching implications of these simple observations are demonstrated through two examples: (i) circular dichroism spectrum deconvolution, and (ii) receptor-ligand affinity determination. These actual comparative examples illustrate the potential errors that can be incorporated into the biophysical parameters of IDPs, due to systematic misestimation of their concentration. This leads to inaccurate description of IDP functions.

High-Throughput Identification of Mammalian Secreted Proteins Using Species-Specific Scheme and Application to Human Proteome.

Secreted proteins are widely spread in living organisms and cells. Since secreted proteins are easy to be detected in body fluids, urine, and saliva in clinical diagnosis, they play important roles in biomarkers for disease diagnosis and vaccine production. In this study, we propose a novel predictor for accurate high-throughput identification of mammalian secreted proteins that is based on sequence-derived features. We combine the features of amino acid composition, sequence motifs, and physicochemical properties to encode collected proteins. Detailed feature analyses prove the effectiveness of the considered features. Based on the differences across various species of secreted proteins, we introduce the species-specific scheme, which is expected to further explore the intrinsic attributes of specific secreted proteins. Experiments on benchmark datasets prove the effectiveness of our proposed method. The test on independent testing dataset also promises a good generalization capability. When compared with the traditional universal model, we experimentally demonstrate that the species-specific scheme is capable of significantly improving the prediction performance. We use our method to make predictions on unreviewed human proteome, and find 272 potential secreted proteins with probabilities that are higher than 99%. A user-friendly web server, named iMSPs (identification of Mammalian Secreted Proteins), which implements our proposed method, is designed and is available for free for academic use at: http://www.inforstation.com/webservers/iMSP/.

In silico approaches for predicting the half-life of natural and modified peptides in blood.

This paper describes a web server developed for designing therapeutic peptides with desired half-life in blood. In this study, we used 163 natural and 98 modified peptides whose half-life has been determined experimentally in mammalian blood, for developing in silico models. Firstly, models have been developed on 261 peptides containing natural and modified residues, using different chemical descriptors. The best model using 43 PaDEL descriptors got a maximum correlation of 0.692 between the predicted and the actual half-life peptides. Secondly, models were developed on 163 natural peptides using amino acid composition feature of peptides and achieved a maximum correlation of 0.643. Thirdly, models were developed on 163 natural peptides using chemical descriptors and attained a maximum correlation of 0.743 using 45 selected PaDEL descriptors. In order to assist researchers in the prediction and designing of half-life of peptides, the models developed have been integrated into PlifePred web server (http://webs.iiitd.edu.in//raghava/plifepred/).

Analysis and prediction of presynaptic and postsynaptic neurotoxins by Chou's general pseudo amino acid composition and motif features

Presynaptic neurotoxins and postsynaptic neurotoxins are two important neurotoxins isolated from venoms of venomous animals and have been proven to be potential effective in neurosciences and pharmacology. With the number of toxin sequences appeared in the public databases, there was a need for developing a computational method for fast and accurate identification and classification of the novel presynaptic neurotoxins and postsynaptic neurotoxins in the large databases. In this study, the Multinomial Naive Bayes Classifier (MNBC) had been developed to discriminate the presynaptic neurotoxins and postsynaptic neurotoxins based on the different kinds of features. The Minimum Redundancy Maximum Relevance (MRMR) feature selection method was used for ranking 400 pseudo amino acid (PseAA) compositions and 50 top ranked PseAA compositions were selected for improving the prediction results. The motif features, 400 PseAA compositions and 50 PseAA compositions were combined together, and selected as the input parameters of MNBC. The best correlation coefficient (CC) value of 0.8213 was obtained when the prediction quality was evaluated by the jackknife test. It was anticipated that the algorithm presented in this study may become a useful tool for identification of presynaptic neurotoxin and postsynaptic neurotoxin sequences and may provide some useful help for in-depth investigation into the biological mechanism of presynaptic neurotoxins and postsynaptic neurotoxins.

A High-Throughput Mutational Scan of an Intrinsically Disordered Acidic Transcriptional Activation Domain.

Transcriptional activation domains are essential for gene regulation, but their intrinsic disorder and low primary sequence conservation have made it difficult to identify the amino acid composition features that underlie their activity. Here, we describe a rational mutagenesis scheme that deconvolves the function of four activation domain sequence features-acidity,hydrophobicity, intrinsic disorder, and short linear motifs-by quantifying the activity of thousands of variants invivo and simulating their conformational ensembles using an all-atom Monte Carlo approach. Our results with a canonical activation domain from the Saccharomyces cerevisiae transcription factor Gcn4 reconcile existing observations into a unified model of its function: the intrinsic disorder and acidic residues keep two hydrophobic motifs from driving collapse. Instead, the most-active variants keep their aromatic residues exposed to the solvent. Our results illustrate how the function of intrinsically disordered proteins can be revealed by high-throughput rational mutagenesis.

Classification of Hot and Cold Recombination Regions in Saccharomyces cerevisiae: Comparative Analysis of Two Machine Learning Techniques

Recombination plays a crucial role generating natural genetic variations. Nevertheless, the recombination does not occur randomly in the genome, but with higher allure in some genomic regions while low in other regions, former are called hot recombination regions and later are called cold regions for recombination. With the advancement of genome sequencing techniques, computational methods are required which can efficiently classify the recombination regions. For this we have developed artificial neural network based model which uses amino acid composition features of DNA sequences. Compositional features were used to incorporate its local or short range sequence order information. High accuracy and sensitivity indicates that this model may become a useful tool for identifying the recombination hotspots. Moreover we compared the performance artificial neural network model with support vector machine model with the target class as hot and cold. We found that compositional features gives good classification result which probably reflect the structural and functional characteristics of hot and cold spots.

Investigation and identification of protein carbonylation sites based on position-specific amino acid composition and physicochemical features

BackgroundProtein carbonylation, an irreversible and non-enzymatic post-translational modification (PTM), is often used as a marker of oxidative stress. When reactive oxygen species (ROS) oxidized the amino acid side chains, carbonyl (CO) groups are produced especially on Lysine (K), Arginine (R), Threonine (T), and Proline (P). Nevertheless, due to the lack of information about the carbonylated substrate specificity, we were encouraged to develop a systematic method for a comprehensive investigation of protein carbonylation sites.ResultsAfter the removal of redundant data from multipe carbonylation-related articles, totally 226 carbonylated proteins in human are regarded as training dataset, which consisted of 307, 126, 128, and 129 carbonylation sites for K, R, T and P residues, respectively. To identify the useful features in predicting carbonylation sites, the linear amino acid sequence was adopted not only to build up the predictive model from training dataset, but also to compare the effectiveness of prediction with other types of features including amino acid composition (AAC), amino acid pair composition (AAPC), position-specific scoring matrix (PSSM), positional weighted matrix (PWM), solvent-accessible surface area (ASA), and physicochemical properties. The investigation of position-specific amino acid composition revealed that the positively charged amino acids (K and R) are remarkably enriched surrounding the carbonylated sites, which may play a functional role in discriminating between carbonylation and non-carbonylation sites. A variety of predictive models were built using various features and three different machine learning methods. Based on the evaluation by five-fold cross-validation, the models trained with PWM feature could provide better sensitivity in the positive training dataset, while the models trained with AAindex feature achieved higher specificity in the negative training dataset. Additionally, the model trained using hybrid features, including PWM, AAC and AAindex, obtained best MCC values of 0.432, 0.472, 0.443 and 0.467 on K, R, T and P residues, respectively.ConclusionWhen comparing to an existing prediction tool, the selected models trained with hybrid features provided a promising accuracy on an independent testing dataset. In short, this work not only characterized the carbonylated substrate preference, but also demonstrated that the proposed method could provide a feasible means for accelerating preliminary discovery of protein carbonylation.

Identification of the core regulators of the HLA I-peptide binding process

During the display of peptide/human leukocyte antigen (HLA) -I complex for further immune recognition, the cleaved and transported antigenic peptides have to bind to HLA-I protein and the binding affinity between peptide epitopes and HLA proteins directly influences the immune recognition ability in human beings. Key factors affecting the binding affinity during the generation, selection and presentation processes of HLA-I complex have not yet been fully discovered. In this study, a new method describing the HLA class I-peptide interactions was proposed. Three hundred and forty features of HLA I proteins and peptide sequences were utilized for analysis by four candidate algorithms, screening the optimal classifier. Features derived from the optimal classifier were further selected and systematically analyzed, revealing the core regulators. The results validated the hypothesis that features of HLA I proteins and related peptides simultaneously affect the binding process, though with discrepant redundancy. Besides, the high relative ratio (16/20) of the amino acid composition features suggests the unique role of sequence signatures for the binding processes. Integrating biological, evolutionary and chemical features of both HLA I molecules and peptides, this study may provide a new perspective of the underlying mechanisms of HLA I-mediated immune reactions.

Thermostability of Proteins Revisited Through Machine Learning Methodologies: From Nucleotide Sequence to Structure

Background: Predicting thermostability of a protein, given its sequence or structure is a challenging job. Predicting which mutations can render mesophilic proteins thermostable is further challenging. A guided approach for the same is still elusive. Prediction can be done on the three hierarchies of protein organization: nucleotide sequence of the gene, primary amino acid sequences and the three dimensional structure of proteins. However it is still unclear that which level leads to a better predictor of protein thermostability and what combination of parameters in each level is responsible for thermostability of proteins. Methods: Present paper addresses this question by testing numerous unsupervised and supervised machine learning methodologies with multitude of prediction parameters to discriminate thermostable and mesostable proteins and to elucidate parameters in each level that lead to protein thermostability. The present method used 33 features of nucleotide sequences, 19 amino acid composition features and 17 structural features. Numerous machine learning approaches: 11 weighting algorithms, 5 unsupervised, 3 supervised learners and 2 lazy modeling algorithms with different parameters were tested and applied on the three datasets. Results: Results showed that amino acid datasets outperforms the nucleotide and structural datasets in generating the best prediction model for protein thermostability. LibSVM with polynomial kernel when applied on amino acid dataset with 10 fold cross validation results in the best prediction model for classifying thermostable proteins with 90.97% accuracy of prediction. Results also show that CG codons are not deterministic of thermostability. Decrease in Gln content, increase in hydrophobic residues and increase in main chain to main chain hydrogen bonds followed by gamma turns and aromatic aromatic interactions are responsible for protein thermostability. Conclusion: Present study shows that studying different datasets can bring about the importance of features in each hierarchy of protein organization which is responsible for protein thermostability. The results support the long drawn conclusion that protein amino acid composition is highly correlated to protein thermostability. Thus further insight into all these features can lead to the development of a better predictor and classifier of thermostable proteins. The results support the long drawn conclusion that protein amino acid composition is correlated to protein thermostability. The study also shows that testing of multitude of criteria is important before reaching at a definitive conclusion. Keywords: Thermostability, codon, amino acid, tertiary structure, predictor, machine learning.