consensus matrix yielded sets of genes that co-clustered in more than 80% of the trials. The genes belonging to the stable clusters were noted, excluded from the data and the procedure was repeated to identify additional clusters. Because high-dimensional data is challenging to partition into definitive clusters, the advantage of this approach is that gene-gene clusters are derived through the tendency of genes to be grouped together under repeated resampling and hence are more robust to the initialization and cluster size thresholds. Transcripts nominated as landmarks through this process were then tested empirically to assess ability to measure levels accurately in the L1000 assay as described in “Probe and primer design for the L1000 assay” and experimental validation as described in the L1000 reproducibility sections below.
