BioCentury
ARTICLE | Company News

National Human Genome Research Institute, Yale University other research news

June 4, 2001 7:00 AM UTC

NHGRI researchers published in Nature Medicine an artificial neural network (ANN) method to classify cancers according to gene expression. Using 23 small round blue cell tumor (SRBCT) samples and 40 SRBCT cell lines, researchers trained the ANN on a set of 6,000 genes. The ANN was able to accurately predict classification of independent samples (12 tumors, 2 normal tissues and 9 cell lines). Of the 6,000 genes monitored, 96 were found to be discriminatory. Of genes specifically expressed in a cancer type, 41 were not previously reported, including 7 ESTs with no known function. The method also could be applied to staging cancer, the authors said. ...