원문정보
초록
영어
Life activity is closely related to the dynamic change of protein. Protein Phosphorylation is one of the most important proc GSVM-Based Proteochemometrics Modeling (PCM) for Prediction of Kinase-inhibitor Interaction within the protein modification after translation. It is found that more than 30% proteins can be phosphorylated. Abnormal protein kinases can lead to diverse diseases, such as cancers. Kinase inhibition is an effective method for disease treatment. However, some inhibitors are able to interact with several kinases that hidden but interesting kinase/inhibitor relationships may be included. Use of multi-targeted mining that select inhibitors act on a group of kinases increases the chance to achieve clinical antitumor activity. Proteochemometrics is a novel technology to predict inhibitor-kinase interactions from the chemical properties of kinase inhibitors which can help design more selective treatment and show better curative effect and low toxicity. This article uses a novel machine learning method called granular support vector machines (GSVM) to correlate the descriptors of kinase inhibitors and kinases to the interaction activities. GSVM develops on the basis of statistical learning theory and granular computing theory and thus provides an interesting new mechanism to address complex classification problems. Compared with other algorithms, GSVM gets better predictive abilities whose q2=0.89.
목차
1. Introduction
2. Methods
2.1. Data Sets
2.2. Inhibitor and Kinase Descriptors Extraction
2.3. Descriptors Principle Component Analysis (PCA)
2.4. GSVM and PCM Model Validation
3. Results and Discussion
3.1. Theoretical Framework for Algorithm
3.2. Theoretical Framework for Algorithm
3.3. Optimal Lags Extraction for ACC Transform Method
3.4. GSVM Predict Novel Kinase-Inhibitor Associations
3.5. Comparison between Different Algorithms
4. Conclusion
References