Importance of Reproducibility in High-Throughput Biology: Predicting Sensitivity to Chemotherapy

Over the past few years, microarray experiments have supplied a lot of information about biological changes associated with various types of cancer. Many studies identify subgroups of patients with particularly agressive disease, so that we know who to treat. A corresponding question is how to treat them. Given the treatment options available today, this means trying to predict which chemotherapeutic regimens will be most effective.

Several microarray studies have provided such predictions. Unfortunately, ambiguities associated with analyzing the data have made many of these results difficult to reproduce. In this talk, we will describe how we have analyzed the data, and reconstructed aspects of the analysis from the reported results. In some cases, these reconstructions reveal inadvertent flaws that affect the results. Most of these flaws are simple in nature, but their simplicity is obscured by a lack of documentation. These flaws have not, however, prevented clinical trials from being initiated.