There are a number of foods which are poisonous when raw but safe to eat when cooked. Are there any which are safe to eat when raw but poisonous when cooked?
Castor beans. When whole and uncooked, the hard capsule prevents it from being digested, but if you were to cook the beans, breaking open this capsule, you’d be able to be poisoned by the ricininside the pulp.
(Castor oil is made from the beans, but has no ricin, and is perfectly safe to consume.)
Overcooking meat generates carcinogenic substances called polycyclic aromatic hydrocarbons, though experts differ on whether consuming them increase your cancer risk significantly or only a bit.
Are the beans themselves actually eaten whole? If so, I don’t think they count for the purpose of my question.
I wondered that, myself. 
Since cooking usually denatures proteins except in some cases like with prions I think the answer would have to be “no”.
AFAIK, they’re not deliberately eaten at all - the ‘eating whole’ thing is advice about accidental ingestion of the seeds of the plant when grown as an ornamental.
Googling led me to an article “On Eating Raw Mushrooms” which makes the following claim:
So we have a somewhat vague claim that certain unspecified edible species of mushrooms are “considered” poisonous when cooked. Can any Doper mycologists confirm this and provide further details?
But the toxins don’t necessarily have to be proteins. Wikipedia’s entry for rhubarb says “Cooking the leaves with soda can make them more poisonous by producing soluble oxalates.” Oxalate isn’t really that poisonous (give you kidney stones, though), and you don’t eat the leaves, but the above illustrates that there are compounds in plants that aren’t proteins and yet are toxic. And some may change to a more toxic form when heat is applied.
Another common example of a non protein plant toxin would be Strychnine, which is an alkaloid.
IANAMycologist, but I think I’m right in saying that most non-edible Lactarius and all non-edible Russula aren’t poisonous so much as variously unpalatable due to peppery or bitter flavours* - so it could be that brining/pickling fixes that, rendering them edible.
(*Despite this apparent ‘rule of thumb’, there really aren’t any safe generalisations when it comes to mushroom picking - careful identification of species is the only truly reliable way)
Found one. Oddly enough, it’s kidney beans. From Here
To be fair, if cooked properly there is no danger, but it’s the closest I can find. Incidentally, the toxin is Phytohaemagglutinin - a lectin protein.
I’ve eaten kidney beans cooked to higher temperatures plenty of times. It must not be much of a toxin, even potentiated five-fold.
The problem is that they were cooked at a sub-boiling temperature; the lectin in question is destroyed if boiled.
Ah, so there’s only a relatively narrow range of dangerous temperatures, then.
There is a joke somewhere in here about my wife’s cooking…
Seriously though, can’t some poisonous compounds be dissolved out of the pits of stone fruits?
I would guess that the possibility that the shell of a bean is damaged in some way would be a concern even swallowing the beans whole. Castor beans are toxic enough that the lethal dose for a human is 4 - 8 beans.
I’m not sure I’d consider castor oil “perfectly safe”, actually. It’s used as a food additive in small quantities, and as a laxative. A large dose of it is nasty:
Well, your one cup of Kidney bean stew cooked at the improper temperature would be as toxic as 5 cups of bean salad, which is to say, not really noticeably toxic at all.
Unless you consider the aftermath on the porcelain throne toxic…