Predicting Subcellular Location of Mycobacterial Proteins
Author Information
Author(s): Rashid Mamoon, Saha Sudipto, Raghava Gajendra PS
Primary Institution: Bioinformatics Centre, Institute of Microbial Technology, Chandigarh, India
Hypothesis
Can a Support Vector Machine-based method accurately predict the subcellular localization of mycobacterial proteins using evolutionary information and motifs?
Conclusion
A highly accurate method has been developed for predicting the subcellular location of mycobacterial proteins.
Supporting Evidence
- The SVM model using PSSM profiles achieved an overall accuracy of 86.62%.
- A hybrid model combining SVM and MEME/MAST achieved an overall accuracy of 86.8%.
- The method predicts a new class of mycobacterial proteins called 'membrane-attached by lipid anchor'.
- The study is the first to develop a dedicated method for predicting subcellular localization of mycobacterial proteins.
Takeaway
The researchers created a computer program that helps figure out where proteins from mycobacteria are located in a cell, which is important for understanding how these bacteria work.
Methodology
The study used Support Vector Machine models trained on amino acid composition and Position-Specific Scoring Matrix profiles, evaluated through five-fold cross-validation.
Potential Biases
The performance of machine learning techniques is biased by the number of proteins in a class used for training.
Limitations
The method performed poorly on secretory and membrane-attached proteins due to the small number of proteins in these classes used for training.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website