The target was right.
Nothing happened.
We take 15 targets from a real drug pipeline and put them straight into a live Perturb-seq API. Plenty of cells, confirmed knockdown — and almost every answer is zero.
The experiment did not fail. The cell was the wrong question. You make that mistake deliberately over nine lessons, then move to the targets this data can answer. You build the Python client yourself.
How you'll learn
You learn the six endpoints of a Perturb-seq server that is actually running. The real contract, not a mock.
You put in 15 pipeline targets and get almost all zeros. Reading that silence is half of this course.
You write the function that sends the request and turns JSON into a table. You use it for the next fourteen lessons.
Turn the knobs for knockdown, cell count and cell state. You see which designs can answer before you run one.