Out−degree

• Infer linear “gene-to-gene ” networks from “extended Spellman ” data (yeast cell-cycles) of expression levels and rates. • Classical underdetermined linear regression problem (6178 genes and 73 measurements). We infer networks from these genes and from a subset of 327 genes proposed by de Lichtenberg et al (2003) to be cell cycle associ-ated, by adding the LASSO constraint (Tibshirani 1996, see also Gustafs-son et al 2004). • We use a biochemichal network consisting of 690 genes (Milo et al 2002) as a reference network. • We perform a posteriori justification of our approach by analyzing the bio-logical aspects of the networks. Some network properties • Broad out-degree distribution (see fig-ure) • Narrow in-degree distribution (de-sired, but probably due to the LASSO) • Excess of transcription factors among the hubs (Gustafsson et al 2004