NCI-60 Cell Line Transcriptomic and Proteomic Profiling

The NCI-60 dataset contained transcriptomic [31 (link)] and proteomic [32 (link)] tables for a collection of 60 cell lines from the National Cancer Institute (NCI-60). The NCI-60 panel included cell lines derived from various cancer types: colon (7 cell lines), renal (8), ovarian (6), breast (8), prostate (2), lung (9) and central nervous system (6), as well as leukemia (6) and melanoma (8). The gene expression profiles used here were generated using an Agilent platform [31 (link)] and downloaded from Cellminer [33 (link)]. Data were log₂-transformed. To facilitate data interpretation and computation, the transcriptomic data were filtered to exclude probes that did not map to an official HUGO gene symbol and to retain only the probe with the highest average value when multiple probes mapped to the same gene, as previously described in [34 (link)]. Gene invariants across all 60 cell lines, corresponding to genes without any effect between cancer types, were also removed. Filtering produced a dataset of 1,433 genes. The NCI-60 proteome table was also downloaded from Cellminer [33 (link)]. Proteomic data were obtained using high-density reverse-phase lysate microarrays [32 (link)]. Data were log₂-transformed and protein abundance levels were available for 162 proteins [32 (link)].