ETA: It sounds like I’m passing off all blame onto my PI, but for fairness it *is* a bad result and I am probably overlooking an important step or I’m just not good with my hands with this specific assay. It is obviously my fault that it isn’t working, but I’m not intentionally skipping steps or not being careful.
The situation you describe happens ALL the time.
You need to learn to "manage up" without being emotionally affected. This is one of the elementary skills required if you want to survive as a research scientist.
Here are the basic steps I recommend.
1) If you truly believe you followed the protocol correctly, the results are non-replicable. This happens all the time. Accept it as a reality.
2) Your boss is scared that his results are caught to be non-replicable.
3) You need to learn to be able to assuage people's fear, and avoid translating it into anger.
4) Non-replication is a technical issue. Either you made a mistake or the protocol itself has a problem that makes it hard to replicate, or the results were wrong or made up.
5) You need to calmly discuss with your boss and say, I'm trying my best for real, however, the results are not working out. The way you are talking to me make me feel like you don't trust me and are angry and are accusing me of being lazy, which I don't think the right way to solve this problem. Why don't we brainstorm to see which steps might be the weakest link, etc.
6) Your boss sounds like an ineffective manager, which is a dime a dozen in academia. However, your job as his trainee is to
train him so that he can become an effective manager later.
Learn to fix interpersonal conflicts in a professional way is a
foundational skill in science. If this is too hard consider getting some coaching.