C4.5: programs for machine learning
C4.5: programs for machine learning
| Hi-index | 0.00 |
The human immune system is a highly complex machinery tuned to recognize specific molecular patterns in order to distinguish self from non-self proteins. Specialized immune cells can recognize major histocompatibility (MHC) molecules with bound protein fragments (peptides) on the surface of other cells. If these peptides originate from virus or cancer proteins, the immune cells can induce controlled cell death. In silico vaccine design typically starts with the identification of peptides that might induce an immune response as a first step. This is typically done by searching for specific amino acid patterns obtained from peptides known to be recognized by the immune system. We propose a new method for deriving decision rules based on the physiochemical properties of such peptides. The rulesets generated give insights into the underlying mechanism of MHC-peptide interaction. Furthermore, we show that these rulesets can be used for high accuracy prediction of MHC binding peptides.