Analysis of the raw DNA sequence through the use of various bioinformatics (computer-based) methods revealed several interesting features of the human genome, including the preliminary finding that our genome contains approximately 30,000Ð 40,000 protein-coding genes. A similar analysis of restricted raw data by a private concern revealed most of the same features. In a companion study that accompanied the publication of the human genome first draft, Lincoln Stein, his CSHL colleagues, and other members of the International SNP Map Working Group identified some 1.4 million single nucleotide polymorphisms (or SNPs) distributed throughout the human genome. The high-density map of human DNA sequence variation will become useful for identifying disease-related genes, for tailoring therapies to those patients who are most likely to respond, and for several other applications. Importantly, this project was jointly funded by a consortium of pharmaceutical companies, private foundations, and the federal government, providing a model for future ways of funding large projects and making the data freely available to the broader scientific community.

To study and manipulate genes for a wide variety of research, diagnostic, or therapeutic purposes, scientists need to determine the precise fine structure of genes against a backdrop of what is frequently a vast and complex genetic landscape. Conventional bioinformatics software fails when it comes to detecting two important features of genes the very first segments of genes, and the nearby "on" switches of genes called promoters.

Michael Zhang and his colleagues have developed a computer program called First Exon Finder (or FirstEF) that is especially good at finding these first segments and "on" switches of genes. The program is tailored toward detecting these features in the human genome sequence, but it is also useful for annotating other mammalian genomes. Although the total number of genes in an organism's genome depends on subtle, nonuniversal definitions of what constitutes a gene, on the basis of his analysis of the human genome using FirstEF, Michael believes that there are 50,000Ð 60,000 fundamental protein-coding human genes and that the pre-liminary estimate of 30,000Ð 40,000 human genes is too low. We will test these computer predictions and have established a joint research initiative involving the laboratories of Michael Zhang, Greg Hannon, and Dick McCombie. These laboratories will experimentally verify the new pre-dicted genes. Greg will incorporate the findings into a project whose goal is to determine the function of human genes that have heretofore not been assigned a function.