[evandijk70]

Researcher working in a related field. The above is just patently false. There are currently no drugs that can do this in lab-animals, let alone make it through clinical trials in time by 2027.

Computation is also not the bottleneck of research, it's the amount of training data, see ICGC https://dcc.icgc.org for the current state of the art.

Remember that each tumor has 3 billion base pairs (A,C,G,T), each of which can be duplicated, deleted, mutated. It can also be affected by methylation. Expression of the genes comprised by these base pairs is also affected by the micro-environment of the cell. Any machine learning model will have far to many parameters for far to few observations, to pick the rare drug that can cure stage 4 cancer (assuming it even exists) out of millions of possible chemical compounds.