Our immune system uses antibodies to neutralize pathogens such as bacteria and viruses. Antibodies bind to parts of foreign proteins with high efficiency and specificity. We call such parts epitopes. The identification of epitopes, namely epitope mapping, may contribute to various immunological applications such as vaccine design, antibody production and immunological diagnosis. Therefore, a fast and reliable method that can predict epitopes from the whole proteome is highly desirable. In this work we have developed a computational method that predicts epitopes based on sequence information. We focus on using local alignment to extract features from peptides and classifying them using Support Vector Machine. We also propose two approaches to optimize the features. Results show that our method can reliably predict epitopes and significantly outperforms some most commonly used tools.
| Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:kth-170658 |
| Date | January 2015 |
| Creators | Zhu, Yunyi |
| Publisher | KTH, Numerisk analys, NA |
| Source Sets | DiVA Archive at Upsalla University |
| Language | English |
| Detected Language | English |
| Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
| Format | application/pdf |
| Rights | info:eu-repo/semantics/openAccess |
| Relation | TRITA-MAT-E ; 2015:51 |
Page generated in 0.0022 seconds