Large-scale in vitro screenings of (potential) cancer drugs are done to access their effectivity depending on the actual transcriptome of a tumor cell line. Databases like GDSC or CCLE supply data of these experiments.
For each drug - cell line experiment they provide an IC50 value: The concentration of the drug needed to kill of 50% of the cancer cells in vitro.
Why is that value interesting? The value simply depends on the toxicity of the compound. In these databases cytotoxic snake venom would probably be the seen as the best cancer drug, but it would not be a great idea to give that to patients.
Cancer drugs work by killing of cancer cells faster than normal cells. Therefore, I think these IC50 values lack normalization/ comparison to the IC50 of healthy cells. Without this I feel the databases are useless.
Where am I wrong?