Prediction of RNA binding sites in a protein using SVM and PSSM profile

Kumar, Manish; Gromiha, M. Michael; Raghava, Gajendra P. S.

doi:10.1002/prot.21677

Cited by 266 publications

(307 citation statements)

References 12 publications

Supporting

Mentioning

301

Contrasting

Unclassified

Order By: Relevance

“…Recently, we developed a method PPRINT for predicting RNAbinding residues (Kumar et al, 2007a). Using PPRINT, RNA-binding residues were predicted for all the proteins in the main dataset and RBPs were assigned with different thresholds.…”

Section: Rna-binding Residue Prediction Based Approachmentioning

confidence: 99%

“…Prediction of RNA-binding amino acids was done using our earlier developed method PPRINT (Kumar et al, 2007a). Each protein was submitted to PPRINT web-server (http://www.imtech.res.in/raghava/pprint) and fraction of RNA-binding amino acids was calculated by dividing the number of binding residues with total amino acid residues in protein.…”

Section: Rna-binding Amino Acid Predictionmentioning

confidence: 99%

“…We have evaluated the performance of similarity search methods BLAST and PSI-BLAST as well as the method developed for predicting RNA-binding residues, PPRINT (Kumar et al, 2007a), to identify the RBPs with specific thresholds. We have observed that although PPRINT achieved very high specificity, the sensitivity was low.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

SVM based prediction of RNA‐binding proteins using binding residues and evolutionary information

Kumar¹,

Gromiha²,

Raghava³

2011

J of Molecular Recognition

Self Cite

129

120

View full text Add to dashboard Cite

RNA-binding proteins (RBPs) play crucial role in transcription and gene-regulation. This paper describes a support vector machine (SVM) based method for discriminating and classifying RNA-binding and non-binding proteins using sequence features. With the threshold of 30% interacting residues, RNA-binding amino acid prediction method PPRINT achieved the Matthews correlation coefficient (MCC) of 0.32. BLAST and PSI-BLAST identified RBPs with the coverage of 32.63 and 33.16%, respectively, at the e-value of 1e-4. The SVM models developed with amino acid, dipeptide and four-part amino acid compositions showed the MCC of 0.60, 0.46, and 0.53, respectively. This is the first study in which evolutionary information in form of position specific scoring matrix (PSSM) profile has been successfully used for predicting RBPs. We achieved the maximum MCC of 0.62 using SVM model based on PSSM called PSSM-400. Finally, we developed different hybrid approaches and achieved maximum MCC of 0.66. We also developed a method for predicting three subclasses of RNA binding proteins (e.g., rRNA, tRNA, mRNA binding proteins). The performance of the method was also evaluated on an independent dataset of 69 RBPs and 100 non-RBPs (NBPs). An additional benchmarking was also performed using gene ontology (GO) based annotation. Based on the hybrid approach a web-server RNApred has been developed for predicting RNA binding proteins from amino acid sequences (http://www.imtech.res.in/raghava/rnapred/).

show abstract

Section: Rna-binding Residue Prediction Based Approachmentioning

confidence: 99%

Section: Rna-binding Amino Acid Predictionmentioning

confidence: 99%

See 1 more Smart Citation

SVM based prediction of RNA‐binding proteins using binding residues and evolutionary information

Kumar¹,

Gromiha²,

Raghava³

2011

J of Molecular Recognition

Self Cite

129

120

View full text Add to dashboard Cite

show abstract

“…3; Wenzel et al 2012). For RNAprotein interactions, bindN (Wang and Brown 2006), bindN+ , Pprint (Kumar et al 2008), and RNAbindR (Terribilini et al 2007) are commonly used tools to predict RNA-binding residues in proteins (Puton et al 2012). Similarly, RAID also merges the predicted RNA-binding residues and scores from these tools.…”

Section: The Predicted Binding Sites and Network Visualizationmentioning

confidence: 99%

RAID: a comprehensive resource for human RNA-associated (RNA–RNA/RNA–protein) interaction

Zhang

Chen

et al. 2014

RNA

View full text Add to dashboard Cite

Transcriptomic analyses have revealed an unexpected complexity in the eukaryote transcriptome, which includes not only protein-coding transcripts but also an expanding catalog of noncoding RNAs (ncRNAs). Diverse coding and noncoding RNAs (ncRNAs) perform functions through interaction with each other in various cellular processes. In this project, we have developed RAID (http://www.rna-society.org/raid), an RNA-associated (RNA-RNA/RNA-protein) interaction database. RAID intends to provide the scientific community with all-in-one resources for efficient browsing and extraction of the RNAassociated interactions in human. This version of RAID contains more than 6100 RNA-associated interactions obtained by manually reviewing more than 2100 published papers, including 4493 RNA-RNA interactions and 1619 RNA-protein interactions. Each entry contains detailed information on an RNA-associated interaction, including RAID ID, RNA/protein symbol, RNA/protein categories, validated method, expressing tissue, literature references (Pubmed IDs), and detailed functional description. Users can query, browse, analyze, and manipulate RNA-associated (RNA-RNA/RNA-protein) interaction. RAID provides a comprehensive resource of human RNA-associated (RNA-RNA/RNA-protein) interaction network. Furthermore, this resource will help in uncovering the generic organizing principles of cellular function network.

show abstract

“…Else prediction technologies include SVM algorithm [18,19], KNN algorithm [20,21], Bayes algorithm [22,23], Genetic algorithm [24] , Neural Networks [25] , etc.…”

Section: Stratified Decision Tree Algorithmmentioning

confidence: 99%

Research on Data Mining Technologies for Complicated Attributes Relationship in Digital Library Collections

Zhao¹,

Niu²,

Peng³

2014

Appl. Math. Inf. Sci.

View full text Add to dashboard Cite

Abstract:We present here the research work on data mining technologies for complicated attributes relationship in digital library collections. Firstly our work and ideology is introduced as the research background of this paper. Digital library evaluation is an important topic in information systems domain. We creatively import data mining technologies into it to get an intelligent decision support. But traditional data prediction algorithm didn't work well. This is the problem which would be solved in this paper. Secondly related preliminary research is introduced. We researched on attributes of digital library collections, proposed a parallel discretization algorithm based on z-score theory, and by the discretization algorithm discovered a complicated condition attribute relation among attributes, it is the reason why traditional data prediction algorithm didn't work well. At last a stratified decision tree algorithm for value prediction about digital collection is put forward as the ultimate solution to solve the problem. Stratified attribute concept is imported in this algorithm. It can expand the selection of splitting attribute in decision tree from flat information to stereoscopic information, eliminate the influence of complicated condition attribute relationship, nested use existing decision tree algorithms, and solve the bottleneck of data mining application in digital library evaluation.

show abstract

Prediction of RNA binding sites in a protein using SVM and PSSM profile

Cited by 266 publications

References 12 publications

SVM based prediction of RNA‐binding proteins using binding residues and evolutionary information

SVM based prediction of RNA‐binding proteins using binding residues and evolutionary information

RAID: a comprehensive resource for human RNA-associated (RNA–RNA/RNA–protein) interaction

Research on Data Mining Technologies for Complicated Attributes Relationship in Digital Library Collections

Contact Info

Product

Resources

About