Ramblings: March 2023
1 April 2023
Here are collected Twitter ramblings for March 2023. This stack has been delayed for a while because I’m trying to figure out how to put these things together while Substack and Twitter are feuding with each other. I think these links will work because I created them before Twitter shut down live linking from Substack. In the future, I’m going to have to screenshot tweets until everyone learns how to play nice in the sandbox again.
This month’s photo of the month is from Rockhouse Hill Sanctuary in Oxford, CT. This place is one of those hidden gems owned by a local land trust that you don’t know about unless you actively seek out such places. Late winter is a great time to hike in Connecticut. There’s a stark beauty in this season.
I’ve often said that the Rule of 5 should be described as the Guidance of 5 instead. It’s a useful mnemonic for playing the odds on oral absorption, but in too many corners of the medicinal chemistry universe, it’s become the law — inscribed on stone tablets. That kind of thinking is too limiting, because it ignores the probabilistic nature of the Rule of 5. There is plenty of life beyond the Rule of 5. Jan Kihlberg in particular has made a big chunk of an academic career out of exploring the limits of oral absorption space beyond the Rule of 5. Orally absorbed PROTACs would be nigh on impossible if the Rule of 5 was really the limit. It clearly isn’t, and medicinal chemists should broaden their horizons a bit.
Although I’ve reserved the lowest circle of hell for “importantly”, all other adverbs are about one level up from that. They’re words that almost never add anything of value to a manuscript, or any other writing for that matter. I’m a believer in the simple test that if you can remove a word from a sentence and it doesn’t change the meaning, then you remove the word.
There’s no “just” doing anything in drug discovery. Anyone who’s practiced it for a while will tell you: drug discovery is trench warfare. It’s a game of inches, where you’re happy to make it alive to the next trench. Anyone promising much more than that is likely coming in from outside the field without the depth of experience to say such things with authority.
There are legitimate differences of opinion on the correct nomenclature here. Some will correctly point out that many other biological posttranslational processes have the -yl stem, to wit: biotinylation, acetylation, farnesylation, etc. But I nonetheless land squarely in the camp of the “yl” in ubiquitinylation sounding a little redundant. I’ve also seen “ubiquitylation” out there, which seems even more wrong to my ear. Count me on Team Ubiquitination. Some time ago I ran a Twitter poll where this was the clear favorite, so there’s also that.
There are a lot of strong feelings out there about testing on animals. For as long as there’s been a Food and Drug Administration though, we’ve been able to say with no ambiguity that getting an Investigational New Drug (IND) application approved — the beginning of human clinical trials — legally required some degree of animal testing. Ethically, having worked in the pharma industry for as long as I have, trust me — you really don’t want drugs going straight into people without some animal testing first. We’re simply not good enough at biology, or chemistry, or anything else to accurately predict how new drugs are going to behave in humans without at least some modeling in animals. We’re already talking about an industry where the failure rate in clinical trials is something like 95%, and that’s with rigorous preclinical testing, including in animals.
And now have a new law “modernizing” the FDA with the sole purpose of removing required animal testing. While I think we can all agree that we should do our best to minimize the amount of animal testing needed to safely bring new drugs to patients, this law seems to have gotten far ahead of reality on the ground. While the FDA can in principle now accept INDs without animal testing, most pharma companies are not going to want to be the first to dip their toes in that water and risk rejection by the agency. Bringing forward an IND is a highly choreographed process that can take years, and is also the part where costs really start to ramp for a company. Why risk all that? The agency itself will likely take a conservative line here, and it’s going to take an awful lot of convincing to get them to go against decades of precedent.
My greatest fear is that charlatan companies that don’t know any better or simply don’t care are going to abuse this new framework to bring a slipshod drug candidate to patients — and that the FDA might let them. Although I’m not a betting person by nature, I’ll put the over/under on a first successful IND without animal data at least a decade from now.
I sense mysterious forces at work here. This whole thing with OCP ranting and then blocking a bunch of people was odd to begin with. So I guess it’s entitled to an odd conclusion too.
Seriously, molecular glue gang: don’t use MGD as your acronym. Beer is the first thing large swaths of America are going to think of.
This is a true story. Our lab didn’t originate this prank; someone brought in a story about it from another institution, but we all agreed it was wickedly good. It was also an important reminder about leaving important things unattended, and also wasting lab resources. If you put a sample in 10 mL of DCM on the rotovap and then go away for an hour to make sure your sample gets “extra dry”, you’re probably a jerk.
This was the beginning of regular career mindfulness for me. It’s a skill that most people don’t innately possess, so has to be learned. At the time, the workload I was under was crushing. I was working 10+ hours a day, plus more time on weekends, just to keep up with my expected workload. In hindsight, I now see that there was far too much on my plate and not enough surrounding infrastructure to support me. I knew I had become deeply unhappy, but lacked the language to articulate why.
Which leads us directly to this short thread, wherein I describe some of the language needed for these kinds of situations.
A book could be written about the supreme overconfidence of tech bros entering areas that they know very little about and thinking they’re going to revolutionize a whole industry. And so many of them come in with companies like Calico (a wholly-owned subsidiary of Alphabet) with an anti-aging focus. This seems to be an obsession of a certain subset of billionaires who want to live forever. I doubt very much that science is going to make enough progress within their founders’ lifespans to meaningfully extend their lives, so they should probably get on with estate planning like the rest of us.
The first offer is rarely the last and final offer, and here I detail what that typically looks like in the pharma sector. A company that has made you an offer is also invested in you, and wants to reel you into the fold. They all have their limits, but you won’t know where they are if you don’t test them.
This is not to say that drugs can’t have micromolar affinity for their primary target. In the much earlier days of phenotypic drug discovery, micromolar affinity for a primary target was a lot more normal — when the primary target was even known. Compounds back then traded much more on their PK profiles. But it does remain true today, as ever, that the risk of polypharmacology goes up as the nominally efficacious concentration goes up.
A subset of this problem is when functional effects in cells attributed to inhibition of a target occur at concentrations much lower (10x or more) than the concentrations nominally needed for inhibition of the primary target in the first place. Stop and think for a second: in what universe does than make any sense? If you’re inhibiting a cancer target at ~5 uM in a biochemical or cell-based assay but then showing antiproliferative effects in those same cells at ~500 nM — that’s a problem, full stop. Uncommon? Nope, you’ll see this kind of crap in just about every issue of every med chem journal. This is one of the easiest bullshit filters you can apply when reading a med chem paper. If you find this kind of thing, stop reading that paper immediately: the conclusions are probably all nonsense.
Another flavor of “it’s not what you know, it’s who you know”. You often join the pharma industry supremely confident that you know it all, then quickly realize that you know nothing. Then you start to learn, but somewhere midway through that journey, you also realize that you have no hope of ever “knowing it all”, or even a fraction of that. The sooner you realize that drug discovery takes a village, and you only succeed by leaning on lots of different people with diverse domain expertise, the better. That’s the moment you become a bona fide drug discoverer. When you unlock the skill to engage others to move a project along, you can fly.
It’s a grim business, drug discovery. In research, we deal with the routine 90-95% failure rate in our daily work. Then layer on top of that a 95% clinical failure rate, which is after we’ve already nominated a drug candidate to the best of our ability. If it was easy, everyone would do it.
See previous post, and it’s easy to understand why. If you follow a discovery project through to clinical candidate nomination, you might spend 5 years on that project. (That’s just a spitballing average and can vary wildly.) A career is ~35 years, maybe 40 at the outside for most industry folks. So if you’re lucky, you might get 7-8 clinical candidates over a career. With a 95% failure rate, you’d need to work on ~20 clinical candidates to have one that makes it to market. So the statistic, while depressing, is correct. Sometimes folks in senior management will burnish their image by listing all the projects they’ve “worked” on that became drugs. But did they work on those projects, or were they in a supervisory role over the people who were in the trenches? See who the inventors are on the patents. Patents don’t lie.
I certainly didn’t expect Chris Murphy to retweet this, especially since I was disagreeing with him!
Closer to 8000 than 7000 as I’m writing this. Thank you for coming along!
This is probably the most consequential editorial I’ve ever read in J. Med. Chem. and is aimed at reducing the reporting of PD and efficacy studies with inadequate PK characterization. Half the tweetorials I’ve written impinge on this topic, so it’s good to see the idea gaining more traction in publications.
More of the same; see previous post.
Ah yes, this was the month of the great Blue Check Crisis. Which I never had, so will never miss. I wouldn’t rule out becoming a Blue subscriber one day, but it would have to come with meaningful identity verification and offer far more tools that would be of use to me than it does today. I don’t see it happening anytime soon.
I do like a good AMA. The questions give me a good sense of what my followers are interested in hearing more about. Listen to your audience!