Main

The Connectivity Map, put together by scientists at the Broad Institute of MIT and Harvard and recently described in Science (Lamb et al., 2006), is a wealth of genomics information that bench researchers will find immediately useful.

This resource contains a collection of gene-expression profiles captured after treatment of cell lines with well-characterized small molecules—164 in the first release—with a broad range of activities. A suite of informatics tools allows users to compare a gene-expression profile of interest to those present in the database. The power of this type of analysis is illustrated in companion papers that appeared in Cancer Cell (Hieronymus et al., 2006; Wei et al., 2006).

The Connectivity Map was conceived with general users in mind. “It was very much a guiding principle,” says Justin Lamb who led the Connectivity Map effort to “make it easy and accessible for people who might not necessarily be experts in genome analysis.”

To appreciate how this new tool can help bench researchers, take, for example, someone who has screened a small-molecule library and found a compound causing a phenotype of interest. “Under the best circumstances,” says Lamb, “you will be presented with a compound, a structure and a name; you've got some stuff in a tube that does something but it doesn't enlighten your research in any way.” Enter the Connectivity Map, which might move things along considerably by providing functional annotation.

The investigator would upload a gene-expression signature for the compound to the Connectivity Map, which—using the ranking algorithms—would generate a list of compounds with the most similar gene-expression signatures, and thus likely to have the same activity. “The advantage for the bench researcher is that an experiment is immediately suggested,” explains Lamb. It would then be fairly straightforward to figure out whether the compound of interest has the activity suggested by the Connectivity Map. An example of this ability to infer functionality by connecting the effect of an unknown small molecule to that of well-characterized compounds is presented in the study by Hieronymus et al.

Another application consists of making a connection between a disease state and the activity of small molecules present in the database as demonstrated by Wei et al. Using the Connectivity Map, this group was able to match the gene-expression signature of a drug-sensitive leukemia to that of rapamycin. When such an approach is successful, it immediately generates a testable hypothesis about pathways involved in the disease state. Again the benefit is in the shortcut to experimentation. Instead of staring at lists of up- and downregulated genes directly or indirectly involved, the Connectivity Map provides you with a very tangible chemical activity. As Lamb puts it, “It circumvents the lists of genes to go from disease to small molecule.” In addition to a new hypothesis to test, defining a chemical activity involved also brings you much closer to therapeutic intervention.

These examples build a very good case for the usefulness of the Connectivity Map, but not all applications will be as successful, and using this resource will require some judgment calls. The results of a search come with a statistical qualifier, but there is no clear threshold yet for a real match, and this threshold will likely depend on the question that is asked. Finding connections between the activities of two chemical compounds may be relatively straightforward, whereas connecting the activity of a small molecule to a complex disease will not be as clearcut.

Although the Connectivity Map has already proven useful, Lamb and his team have plans for improvements. Of high priority is expansion of the database to include the 1,400 US Food and Drug Administration–approved drugs. This well-characterized set of compounds recapitulates a lot of different chemical activities. As Lamb points out, “It is the distillation of hundreds of years of medicinal chemistry.” The team also appears eager to receive feedback from users to plan further developments.