As a medicinal chemist, I am absolutely full of opinions about any given list of chemical structures. Show me a bunch of screening hits, and I’ll immediately start triaging. We all do. It’s how we’re built. We say this one looks fine, nothing wrong with that one, worked on some before that look like that one over there, don’t like that one, really don’t like that one, never saw a structure like that one over there in my entire career, or we’ll go through the whole pile and sort things out for you.

Or will we? There have been many attempts over the years to quantify this sort of thing, because the way we tend to do it now is, frankly, a bit mystical. “Let the chemists look it over” is fine, as far as it goes, but it’s pretty far from a well-controlled process. That becomes clear when you try to examine it in more detail. First off, try asking one of us why we don’t like a particular compound. Sometimes the answer will be that we worked on something very close to it before, and that the project crashed and burned, and that we’re not going to put ourselves through that again. Fine, but did the project wipe out because of that chemical matter? Or was it the target itself? That’s not always an easy question to answer.

And sometimes the answer is that we recognize some functional group on the molecule that we think is steering it towards trouble. That’s a valid response, because there really are things—rhodanines, polyphenols—that show up more often than you’d think from big screening campaigns, but turn out to be, most of the time, impossible to really go on with. I have no objection to a chemist who sees a hydroquinone structure in a screening hit and crosses it off the list, because I, too, have plenty of objections to hydroquinones as drug candidates. But I’m afraid that we often shade this category over from “Things we have good reason to fear” to “Things we don’t recognize, and therefore fear by default.”

I’ve done this myself, and probably more times than I realize. The problem with this precautionary approach is that it surely eliminates too many useful compounds. A glance through the structures of the best-selling drugs on the market will show a number of species that, to be honest, aren’t too appealing. And some of them go beyond that. Metformin might be the best example, one that stands as a rebuke to medicinal chemists everywhere. It’s a small, strange-looking thing, full of nitrogens, and of a compound class that few chemists have ever spent any time exploring. I have no doubt that if you removed my knowledge of metformin from my brain somehow, that I would surely put a big red X through it the next time it showed up as an assay hit. Every chemist has a line somewhere between, “I’ve never worked with those, but they might be OK” and “I’ve never worked with those, but they look screwy to me.” And metformin, for most of us, is probably on the wrong side of that line. But it appears to be a great drug—it’s treated millions of patients while bringing in billions of dollars, and is still prescribed heavily so many years after first coming to market. Any of us should be so lucky as to produce a drug with that much impact.

There’s an even bigger problem with our med-chem intuition, though. It should more properly be called our med-chem intuitions, because the more closely you examine the situation, the more you start to feel that every chemist is carrying around his or her own list. The compounds we don’t object to are often the sorts of things we can all agree on. But when you go to the ones that get rejected, things aren’t so neat any more. We’ve all been burned by different compound classes over the years, and have thus ended up with different immune systems. We have varying tolerance to odd-looking molecules, and we classify different-looking things as odd. The more chemists you run your list of hits past, the more confused you’ll probably be. And by the end of the process, what you’ll be left with are probably the absolute safest, most boring structures, because those are the only ones that everyone could agree on. All the more interesting stuff will have been trimmed off by one person or another.

This is not the sort of news that upper management likes to hear about. Chemistry is supposed to be able to sort such issues out; that’s their job. Coming back and saying that well, there are a lot of interesting compounds on there, some of which we don’t know a lot about, but we should prosecute all of them, because you never know, that’s not the sort of feedback some managers are looking for. They’d rather have a firm, decisive, “These are good and these are bad”, but that’s often not an honest picture of the world. Not that all higher-level managers are always interested in an honest picture of the world at all times, but that’s a different topic.

It gets worse. A study a few years back looked at this sort of compound-picking exercise, sending various lists through a whole panel of medicinal chemists for approval. This confirmed the varying tastes problem—the compounds that were rejected tended to scatter out among the rejecters themselves. But the organizers of this bake-off also tried sending the same lists of compounds back around to the same chemists, after a cooling-off period of some months. None of them apparently noticed that they were getting repeats, which is interesting. Even more interesting is the fact that when they marked up the lists a second time, they accepted and rejected different compounds than before. In fact, the variation between chemists was just about the same as the variation between the same chemist’s picks, at two different times.

Well, that’s embarrassing, to be sure. But it’s also believable. I’d like to think that I would have done better, that I’d have spotted that they were running a similar list back on me, but I honestly doubt that I would have, especially after some time had passed. One sees an awful lot of compound structures in this business. And I’d like to imagine that my picks would have been consistent, either way, but doubtless I’m just flattering myself there, too. The best I could hope for is that my own preferences, assayed two different times, might be a bit more consistent than asking someone down the hall to do it the second time, but that’s not exactly a glorious standard to aspire to, is it?

So in general, I think that we chemists should probably loosen up a bit when it comes to what structures we’re willing to work with. I’m not advocating advancing everything—I’m still not going to work on a phenolic rhodanine, sorry—but we’re probably tossing out perfectly valid lead compounds in the process of tossing out the nasties. We should, I think, just admit that chemical intuition, though valuable, is not an exact science, and swallow our pride a bit. There’s plenty more pride where that came from, I’m pretty sure.