- Poster presentation
- Open Access
Accurate prediction of BRCA1 and BRCA2 heterozygous genotypes using expression profiling of lymphocytes after irradiation-induced DNA damage
© BioMed Central Ltd 2008
- Published: 13 May 2008
- Support Vector Machine
- Mutation Carrier
- Support Vector Machine Classifier
- Microarray Technology
- BRCA2 Gene
Germline mutations in BRCA1 and BRCA2 genes predispose women to an increased risk of breast/ovarian cancer. Both genes have important roles in DNA damage repair and are implicated in gene expression regulation. We have previously shown that normal fibroblasts from mutation carriers can be distinguished from noncarriers following radiation-induced DNA damage. In this new study we used lymphocytes to determine whether these also show differential response to induced DNA damage and whether expression profiling using microarray technology could be used to accurately predict the BRCA genotype.
Short-term lymphocyte cultures were established from fresh blood samples from 20 BRCA1 and 20 BRCA2 mutation carriers and from 10 negative controls (individuals tested negative for the mutation present in the family). Lymphocytes were subjected to 8 Gy ionizing irradiation to induce DNA damage and RNA was extracted 1 hour post γ-irradiation. For expression profiling, genome-wide (30 K) spotted cDNA microarrays manufactured by the Cancer Research UK Microarray Facility were used. We then applied the support vector machine (SVM) classifier with statistical feature selection to determine the best feature set for predicting BRCA1 and BRCA2 heterozygous genotypes. We also investigated the prediction accuracy using a nonprobabilistic classifier (SVM) and a probabilistic classifier (Gaussian process classifier).
We achieved high accuracy (92% to 96%) in predicting the mutation carrier status. We shall present the detailed outcome of using the SVM classifier and the Gaussian process classifier in the task of distinguishing between the three classes, BRCA1 and BRCA2 mutation carriers and noncarriers, and evaluate whether this microarray technology can be used to facilitate the clinical detection and classification of mutations.