Transcription profiling of human set of heterogenous ovarian tumor samples
Ontology highlight
ABSTRACT: We have previously described differences in the expression profiles of tumors by comparing tumors of two known classes: low malignant potential versus highly malignant (invasive). This report presents an effort to discover unknown classes within a large heterogeneous set of ovarian tumors, using unsupervised learning approaches. We were able to define four classes in a set of 74 ovarian cancer tumors. Groups identified as A1 and A2 were closely related and correlated with the âinvasiveâ pathology class while the group identified as B2 was correlated with low malignant potential tumors; group B1 consisted of a mixture of low potential and invasive samples. We selected characteristic candidate genes, which were validated by quantitative-PCR and by comparison to other published studies. Experiment Overall Design: We used unsupervised classification methods to identify unknown classes in a set of 74 cancer samples
ORGANISM(S): Homo sapiens
SUBMITTER: Jaroslav Petr Novak
PROVIDER: E-GEOD-6822 | biostudies-arrayexpress |
REPOSITORIES: biostudies-arrayexpress
ACCESS DATA