profiles is assessed and quantified by a normalized score, from 1 for a molecule that reverses the signature to 1 for a molecule which induces gene expression changes Celecoxib Celecoxib Celebrex Celebrex similar to the query signature. Our strategy was to query the Connectivity Map with a list of genes differentially expressed in infected cells to find molecules that induced the opposite gene expression changes. We hypothesized that such molecules may influence host cell metabolism in such a way that effective viral replication would be altered. A critical step in this screening was to define the query signature. As the number of upregulated genes was very low in the list of 300 genes defined by the analysis, a lack of specificity resulting from a loss of information for up regulated genes could be introduced in drug selection if the signature was not corrected for this bias.

By selecting genes with the most drastic changes in level of expression, we were able to define a signature of 20 genes for influenza A virus infection with similar amounts of those up and down regulated. By querying the connectivity CYT997 map with this concise signature, we obtained c scores for 6100 instances, representing more than 1000 molecules in various conditions. We selected CYT997 those associated with the most strongly anticorrelated signatures and which had a p value less than 0.5%.
Applying this filtering step left us with eight candidate molecules: harmol, rilmenidine, brinzolamide, ribavirin, calcium folinate, 2 aminobenzenesulfonamide, merbromin and midodrine.
The relevance of our selection was supported by the fact that ribavirin, an already known influenza virus inhibitor, was identified with a negative enrichment of 0.83 and a pvalue of 0.00157. Except for the topical antiseptic merbromin, the other selected molecules have various therapeutic indications but are not referenced as antivirals. Graphs in Figure 5C report how the different genes of the infection signature behave in the expression profile of the selected molecules.
Although the genes down regulated during infection are generally up regulated in response to the molecule and conversely the up regulated genes of the signature are globally down regulated by the molecule, none of the molecules available in this data bank were able to completely reverse the infection signature.
3 Evaluation of the antiviral potency of the selected drugs on H3N2 viral growth We assessed the effect of the eight selected molecules on influenza replication in vitro. Cell viability, as assessed by the neutral red assay, and viral growth, as quantified by a neuraminidase activity test, were conducted in parallel. Before using the NA activity test as an indirect measurement for viral impairment, we checked firstly that the different influenza viruses used in this study had sufficient neuraminidase activities to be quantified using this method. For all tested viruses and for a signal to background ratio between 2 and 70, the fluorescence was proportional to the amount of virus present. During the evaluation of the drug panel, all signal to background ratios were included between 2 and 70. Secondly, we controlled that the different molecules did not inhibit the enzymatic activity of NA to be sure that a drop in RFU would only reflect a drop of viral titer. While conce