Hacker News Reader

I only skimmed the article, but the entire thing seemed like some fluff to make a connection between machine learning and biology research.

Cancer and genetic instability has been known for a very long time (hallmarks of cancer). Many energy/maintenance pathways are impaired or outright broken in cancer cells. Despite this knowledge, effectively targeting these “weaknesses” is far from trivial. Broken pathway does not a cancer cure make.

View on HN

Which makes sense. Chemo (limiting cell division, etc.) and/or starvation are "nuclear" options which rely on healthy cells weathering stress more so than cancer cells. Targeting (essentially assumed-to-be-novel) cancer cells specifically (say oncolytic viruses, pathway-blocking molecules or antibodies) seems extremely difficult without sheer luck of prior research matching particular tissues.

But they actually did effectively target them. Six out of six complete remissions.

Reply on news.ycombinator.com