It is now possible to create large numbers of structurally diverse small molecule compounds. However, many compounds fail to engage specific protein targets or to even enter living cells. We asked whether an L1000 profile could serve as a sensor for biological activity. If so, screening chemical libraries with L1000 might enable rapid elimination of compounds lacking obvious activity and help prioritize others for subsequent cell-based screening. Consistent with our earlier studies (Wawer et al., 2014), we found that whereas 2,232/2,429 (92%) established drugs yielded a strong L1000 transcriptional response (defined as Transcriptional Activity Score (TAS) >0.2; see STAR Methods), only 2,418/16,527 (15%) un-optimized compounds had high TAS scores. We note, however, that compounds with cell-type selective bioactivity might be missed by this approach.
