The paper must be within 4 pages, including title, author's name, affiliation, address and liaison author's phone, fax, email

propose a novel approach, which is tive matrix factorization (NMF), to ide expression data. One of NMF it can directly process these data tion. Firstly, we design an optimal F approach. Compared with the rithms, our algorithm is more stable ry fast. We have coded the final hly optimized C. Secondly; we of NMF in the extraction of the erns from genome wide expression e simulation experiments are made he efficiency of NMF algorithm. our at NMF can be used as a powerful biologically meaningful expression mic wide expression data.