High-accuracy prediction of bacterial type III secreted effectors based on position-specific amino acid composition profiles

Motivation: Bacterial type III secreted (T3S) effectors are delivered into host cells specifically via type III secretion systems (T3SSs), which play important roles in the interaction between bacteria and their hosts. Previous computational methods for T3S protein prediction have only achieved limi...

Full description

Saved in:
Bibliographic Details
Published in:Bioinformatics Vol. 27; no. 6; pp. 777 - 784
Main Authors: Wang, Yejun, Zhang, Qing, Sun, Ming-an, Guo, Dianjing
Format: Journal Article
Language:English
Published: Oxford Oxford University Press 15.03.2011
Subjects:
Algorithms
Amino Acid Sequence
Amino acids
Bacteria
Bacterial Proteins - chemistry
Bacterial Proteins - classification
Bacterial Secretion Systems
Bayes Theorem
Bioinformatics
Biological and medical sciences
Composition effects
Computational Biology - methods
Effectors
Fundamental and applied biological sciences. Psychology
General aspects
Genome, Bacterial
Mathematical models
Mathematics in biology. Statistical analysis. Models. Metrology. Data processing in biology (general aspects)
Models, Statistical
Proteins
Ralstonia solanacearum
Ralstonia solanacearum - chemistry
Secretions
Sequence Analysis, Protein - methods
Software
Support vector machines
Composition
Accuracy
Position
Secretion
Aminoacid
Prediction
Bacteria
Profile
ISSN:1367-4803, 1367-4811, 1367-4811, 1460-2059
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Motivation: Bacterial type III secreted (T3S) effectors are delivered into host cells specifically via type III secretion systems (T3SSs), which play important roles in the interaction between bacteria and their hosts. Previous computational methods for T3S protein prediction have only achieved limited accuracy, and distinct features for effective T3S protein prediction remain to be identified. Results: In this work, a distinctive N-terminal position-specific amino acid composition (Aac) feature was identified for T3S proteins. A large portion (∼50%) of T3S proteins exhibit distinct position-specific Aac features that can tolerate position shift. A classifier, BPBAac, was developed and trained using Support Vector Machine (SVM) based on the Aac feature extracted using a Bi-profile Bayes model. We demonstrated that the BPBAac model outperformed other implementations in classification of T3S and non-T3S proteins, giving an average sensitivity of ∼90.97% and an average selectivity of ∼97.42% in a 5-fold cross-validation evaluation. The model was also robust when a small-size training dataset was used. The fact that the position-specific Aac feature is commonly found in T3S proteins across different bacterial species gives this model wide application. To demonstrate the model's application, a genome-wide prediction of T3S effector proteins was performed for Ralstonia solanacearum, an important plant pathogenic bacterium, and a number of putative candidates were identified using this model. Availability: An R package of BPBAac tool is freely downloadable from: http://biocomputer.bio.cuhk.edu.hk/softwares/BPBAac. Contact:  djguo@cuhk.edu.hk Supplementary information:  Supplementary data are available at Bioinformatics online.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ISSN:1367-4803
1367-4811
1367-4811
1460-2059
DOI:10.1093/bioinformatics/btr021