PreMieR is a computational method developed for predicting mannose interacting residues (MIRs) in proteins using local composition-based machine learning approaches.
Mannose-binding proteins (MBPs) are important components of the innate immune system. These proteins recognize mannose molecules present on pathogens and help activate immune defense mechanisms such as complement activation and phagocytosis.
This study introduces Support Vector Machine (SVM)-based prediction models using:
- Binary profile patterns
- Evolutionary information (PSSM)
- Composition profile of patterns (CPP)
Among these, composition-based models achieved the best performance.
Identification of Mannose Interacting Residues Using Local Composition
- Sandhya Agarwal
- Nitish Kumar Mishra
- Harinder Singh
- Gajendra P. S. Raghava
PLoS ONE
2011
Volume 6, Issue 9
e24039
https://doi.org/10.1371/journal.pone.0024039 https://github.com/Piyushh1104/PreMier
This study presents machine learning models for predicting mannose interacting residues in proteins.
A dataset of mannose-binding proteins was created from Protein Data Bank (PDB) structures. Different SVM-based approaches were evaluated using:
- Binary profiles
- PSSM evolutionary profiles
- Composition profile patterns
The composition-based approach achieved the best performance with:
- MCC up to 0.74
- Accuracy up to 86.64%
The study also developed a web server and standalone software called PreMieR.
Mannose-binding proteins are involved in:
- Innate immunity
- Pathogen recognition
- Complement activation
- Opsonization
- Phagocytosis
These proteins recognize mannose molecules present on microbial surfaces.
- Protein Data Bank (PDB)
- SuperSite documentation
- 120 non-redundant mannose-binding protein chains
- Less than 25% sequence similarity
- 1029 mannose interacting residues
- 38136 non-interacting residues
Patterns were encoded into binary vectors.
- MCC: 0.19
- Accuracy: 59.60%
Evolutionary information was incorporated using PSI-BLAST generated PSSM profiles.
- MCC: 0.32
- Accuracy: 65.66%
Patterns were represented using amino acid composition.
- MCC: 0.74
- Accuracy: 86.64%
This method significantly outperformed binary and PSSM approaches.
The study found that certain amino acids are preferred in mannose interaction:
- Aspartic acid (D)
- Glutamic acid (E)
- Asparagine (N)
- Glutamine (Q)
- Serine (S)
- Threonine (T)
- Tryptophan (W)
- Tyrosine (Y)
These residues contribute to carbohydrate recognition and binding.
Different pattern lengths were tested:
- 17
- 19
- 21
- 23
- 25
Best performance was obtained using:
- Window size 23
- MCC: 0.74
The models were evaluated using:
- Sensitivity
- Specificity
- Accuracy
- Matthews Correlation Coefficient (MCC)
- Area Under Curve (AUC)
The prediction method was implemented as an online web server.
- Predict mannose interacting residues
- Adjustable thresholds
- Graphical visualization
- Composition-based prediction
http://www.imtech.res.in/raghava/premier/
PreMieR can be useful for:
- Protein-carbohydrate interaction analysis
- Immunology research
- Host-pathogen interaction studies
- Functional annotation of proteins
- Drug discovery
- Glycobiology research
- Support Vector Machine (SVM)
- PSI-BLAST
- Position Specific Scoring Matrix (PSSM)
- Composition Profile of Patterns (CPP)
- Machine Learning
- Bioinformatics
This study demonstrates that local amino acid composition can effectively predict mannose interacting residues in proteins.
The CPP-based approach performed substantially better than binary and evolutionary profile-based methods.
The work provides:
- A benchmark dataset
- A prediction framework
- A publicly accessible web server
- Insight into protein-carbohydrate interactions
Agarwal S, Mishra NK, Singh H, Raghava GPS (2011)
Identification of Mannose Interacting Residues Using Local Composition.
PLoS ONE 6(9): e24039.
DOI: https://doi.org/10.1371/journal.pone.0024039
Email: raghava@iiitd.ac.in
Address:
Indraprastha Institute of Information Technology Delhi
This project/documentation is intended for academic and research purposes only.