A simple and robust method for connecting small-molecule drugs using gene-expression signatures

TL;DR

This paper introduces a simple, robust method for connecting small-molecule drugs via gene-expression signatures, improving specificity and sensitivity over previous approaches, and aiding drug development and toxicology assessment.

Contribution

The authors propose a new gene-expression profile construction and connection scoring method that includes statistical significance evaluation, outperforming the original Connectivity Map.

Findings

Successfully identified anti-estrogen effects of raloxifene and tamoxifen

Achieved higher specificity and sensitivity than previous methods

Potential applications in drug development and toxicology

Abstract

Interaction of a drug or chemical with a biological system can result in a gene-expression profile or signature characteristic of the event. Using a suitably robust algorithm these signatures can potentially be used to connect molecules with similar pharmacological or toxicological properties. The Connectivity Map was a novel concept and innovative tool first introduced by Lamb et al to connect small molecules, genes, and diseases using genomic signatures [Lamb et al (2006), Science 313, 1929-1935]. However, the Connectivity Map had some limitations, particularly there was no effective safeguard against false connections if the observed connections were considered on an individual-by-individual basis. Further when several connections to the same small-molecule compound were viewed as a set, the implicit null hypothesis tested was not the most relevant one for the discovery of real connections. Here we propose a simple and robust method for constructing the reference gene-expression profiles and a new connection scoring scheme, which importantly allows the valuation of statistical significance of all the connections observed. We tested the new method with the two example gene-signatures (HDAC inhibitors and Estrogens) used by Lamb et al and also a new gene signature of immunosuppressive drugs. Our testing with this new method shows that it achieves a higher level of specificity and sensitivity than the original method. For example, our method successfully identified raloxifene and tamoxifen as having significant anti-estrogen effects, while Lamb et al's Connectivity Map failed to identify these. With these properties our new method has potential use in drug development for the recognition of pharmacological and toxicological properties in new drug candidates.