🧵 Hi new followers! I’m Keith, medicinal chemist
@ArvinasInc
. I post on targeted protein degradation, med chem “wisdom”, industry careers, and various pharmacology/PK tweetorials. This thread of threads (!) rolls up links to all of my previous tweetorials; see below.
My son texted me a bit upset this morning because he broke a beaker in his high school chemistry lab and brother can all we chemists tell him some stories
What’s the biggest / most expensive piece of glassware you’ve ever broken?
There’s a frequent perception among recent entrants to drug discovery world that if we could just find drug candidates faster, this will be a useful accelerant to the overall process. The issue is: this thinking is mostly (but not entirely) incorrect. 🧵 1/
Jensen Huang’s comments from JPM24 are now summarized in a Nvidia blog post. He doubles down on exactly the kinds of things that I’ve been cautioning are wildly optimistic. More below. 👇🏽 1/
Time for a pharmacology tweetorial (of sorts) on one-eyed sheep. It’s an odyssey that began on Idaho ranches in the 1950s and ended over half a century later with the approval of several new cancer drugs for basal cell carcinoma and AML. 1/
❗️BREAKING at
@AACR
#Targets21
: Mirati Therapeutics has disclosed the structure of MRTX1133, their reversible KRAS G12D inhibitor. Fantastic story. Plus hints of a new pan-KRAS inhibitor. Congratulations!
@ChrisMurphyCT
I suggest checking in with some actual chemists (like me) on this. ChatGPT has only a superficial and often wrong grasp of chemistry. It’s only convincing to lay people because it’s a great BS generator. Same logic to many other things it’s been repped to “understand”.
Some personal news:
I’m happy to share that I’ve been promoted to Vice President, Discovery Chemistry at
@ArvinasInc
Huge thanks to the company & its leadership for entrusting me with this new role and to all the folks who have taught and mentored me along the way. ☺️
If you read one thing today, make it this devastatingly moving editorial from Donald Weaver. I’ve heard many stories of how folks became true believers in the drug discovery cause & the power of medicinal chemistry, but this one tops them all. Big feels.
Time for another pharmacology tweetorial, this time about the log-normal distribution and its importance to the interpretation, and aggregation, of assay data in drug discovery. 1/
A little more on the ongoing attempts of the tech folks to reduce biology to its constitutive components. While I’ve said this attempt at reduction may be impossible, it’s actually the converse — reconstituting the whole from the parts — where the problem lies. More follows👇🏽 1/
Time for another pharmacology tweetorial, this time on structural alerts (part 1: redox cyclers) – or broadly speaking, why medicinal chemists don’t like to work with certain functional groups if they can help it. 1/
Time for another pharmacology tweetorial, this time on irreversible inhibitors, their biochemical characterization, and their chemical optimization. 1/
Today I talked to the spouse of a cancer patient who is doing well on dabrafenib + trametinib combo — the former of which I helped to discover — and readers, it was one hell of an emotional moment 😭
This is why we do the thing.
Since we’re all watching The Last of Us, is it really true that we have nothing (other than bombs) for the coming fungal pandemic? Time to review what we’ve got in the pharmacopeia for fungal infections. Lest we all turn into zombies. 🧵 1/
Pretty sure I’m going to sleep okay tonight despite this clearly career-ending McKinsey white paper.
For those who need to hear it again: generating “potential molecules” is not the rate-limiting step in drug discovery.
Y’all remember this the next time someone shows you their x-ray crystal structure and treats it as “the truth”. They are snapshots only. Useful snapshots, but snapshots nonetheless.
Many words over the weekend about pay in academia. All I can say is, over here in industry, we pay well and give you space to have a life. I encourage folks to consider all options. "Academia or bust" feeds a false narrative that an academic job is the only job worth having. 1/2
Once again Derek saved me the trouble of wading through overstated stats on clinical success rates of “AI-discovered” candidates — using that term in its loosest sense. The breakdown on “AI-discovered” targets, meanwhile, is nothing short of comical.
Time for another pharmacology tweetorial, this time about the distinction between clearance and intrinsic clearance. And also a cautionary tale about how we’re so often misled by over-focus on the former. 1/
Clinical trials are the rate-determining (and cost-determining) step of the drug R&D process. Yet we have so much effort focused on innovation for all the non-rate limiting steps.
The chemists can tell you what the reaction rate enhancement from that will be.
I say this with some regularity, and this kind of thing underscores it: academia and industry have different value systems. We don’t care so much about your publication prestige. We don’t even necessarily care about publications at all! 1/2
Time for another pharmacology tweetorial, this time a second installment on structural alerts. We’ll cover more things about anilines and phenols beyond redox cycling. We’ll also talk about glucuronides generally and a reactive subset (acyl glucuronides) specifically. 1/
Folks, for the love of god, please stop running your in vitro studies with your compound concentration at 20 uM. This is just not credible stuff, and yet I run across papers almost daily that do it.
I wouldn’t trade in my pharma/biotech industry career, despite the churn. You don’t worry much about funding, get to do cutting edge science, get to directly impact human health. If you’re a pragmatist, it’s the place to be. Control of your research is kind of an illusion anyway.
This is a quite misleading credit grab imo. Penn treated Dr. Karikó poorly while she was there, pulling her off tenure track to adjunct status in 1995 and not reinstating her in 2013 after the key Nobel papers were published. Bad judgment then, backpedaling now.
Continuing the Friday pharma industry
#EarlyCareer
tip series. Last time we talked about phone interviews. This time let’s talk about the next step, a really big one: on-site interviews. 🧵 1/
If you read one pharmacology paper in January, make it this one that was sent my way by
@pwk2013
. Although it speaks about the difficulty of measuring intracellular free drug concentrations, it’s largely a defense of the free drug hypothesis. 1/6
For drug discovery scientists, I routinely admonish folks to look at the dose-response curves and not just settle for the summary 4-parameter fits like an IC50. Often there’s a great deal of texture lurking just below the surface!
If there was only one scientific practice I could teach to every scientist regardless of stage or field I think it would be: look at the data. Spot check it. Find a few data points and trace them through to see if they make sense. Look at the raw data. Don't just do analyses.
Good context from Derek re: AlphaFold 3, to which I’d add: remember that the training set is king. AlphaFold is built on decades of deposited structural data, and there’s nothing else of remotely comparable quality for modeling other biological phenomena.
“My philosophy is that you cannot make new discoveries behind the office desk, you make them in your fume hood.”
Chemistry laureate Morten Meldal talks about his love of science and making discoveries in his official Nobel Prize interview. Watch here:
I’ve taken out >1 vacuum manifold - the ones with the threaded stopcocks are a wee bit prone to over-tightening in my hands 😬
And in so doing learned the value of a good in-house glassblower
We had someone leave their rotovap running over lunch. So we removed & safely set aside the flask, found an identical clean flask, and dropped it into the water bath. Left the vap spinning with bump trap on. The look of horror when they got back from lunch was priceless. 😂
Hello
#chemtwitter
- do you know of any good chemistry April Fool's pranks or stories? I'm thinking of writing an upcoming
#Newscripts
column about this for
@cenmag
. Please RT.
I’m reminded that today is the 11th anniversary of the approval of dabrafenib (Tafinlar). Very emotional day. How fortunate I was to be part of a team that discovered a marketed drug, and how far we still have to go in cancer treatment. Working hard every day to make it happen!
Favorite quote. Those of us in industry working on treatments for any disease know this brutal truth all too well. It’s why we get on our high horse about reproducibility and data quality.
This actually just came up today, but: most drug discovery scientists will go their entire careers without working on a team that discovers a drug. Which is nominally what we’re hired to do. 💊
The chemistry Nobel announcement is almost upon us, which means it’s time for the annual day-before tradition.
Who will win this year? Wrong answers only.
I don’t post too many
#ProudDad
moments because my kids get annoyed when I do, but I got an exemption for this one. My oldest just completed his board of review and joined his old man as an Eagle Scout! Congrats kiddo! 🦅
If you need a reminder that static protein structures aren’t ground truth, here you go. Does a great job highlighting some challenges any AI/ML drug discovery approach is going to face.
h/t
@pwk2013
for sending my way
Can’t emphasize enough for folks who are new to crafting slide decks in pharma/biotech: less is more. Lead with key points, go to details if time permits. Convey with graphics & minimize text. If you’re in 10-12 point font, you’ve lost the plot. Leave ample time for discussion.
A big difference between tech vs pharma is the “move fast and break things” mentality. Pharma can do this up to a point, but eventually the “things” are people. Then ethics and patient safety and regulation kick in. Not sure this has a tech equivalent 🤔
Continuing the occasional Friday pharma
#EarlyCareer
series. Last time we talked about negotiating a job offer. This time let’s talk about what to do when you don’t get a job offer. 🧵 1/
Continuing the occasional Friday pharma
#EarlyCareer
series. Last time we talked about site interviews. This time let’s talk about negotiating job offers. (Next time we’ll deal with the inevitable rejections.) 🧵 1/
Monday rant: significant figures. Folks, please don’t report an IC50 as 5.027 nM in a paper. There’s no way your measurements are that precise. This is what happens when you blindly take curve fit output from software without thinking. For most assays, 2 sig figs is about right.
Subtweet
You don’t have to have your field of study and expertise all sorted out when you’re a kid.
Learn lifelong. Pick up new skills. Move yourself in a new direction whenever it suits you, and don’t let anyone tell you you’re too old to pick up the new thing.
Life is short.
Total synthesis: still relevant training for medicinal chemists? Yes, as this Viewpoint lays out. Medicines are largely organic molecules. Understanding their structure and reactivity is prerequisite to being a good medicinal chemist. However… 1/
Time for another pharmacology tweetorial on another layer of the selectivity problem: drug promiscuity. Both the things we can control, and the things that are out of our hands thanks to good ol’ Mother Nature. 1/
Drug Discovery Axiom 8: if you have to take your compound concentrations into the micromolar range to see a pharmacological effect, there’s a high chance said pharmacological effect is bullshit.
You can rule the world of off-target effects at 10 uM, or even less.
Though showing some wear and tear after an unkind two decades of house moves and job changes, I still have that sugar packet today. I’ll never forget that day. 10/10
For this
#SynthesisSaturday
, it’s gotta be the reaction that launched a thousand syntheses in chemical biology: the OG description of the SPAAC reaction from the Bertozzi lab. Amazing how much content they packed into a 2 page JACS communication.
Read enough papers that you can post something about a paper ~daily here on the bird app. This is a win-win that will grow your audience, your mind, and your reading discipline. It’s served me well over the years.
New in
@ACSBioMed
J. Med. Chem.: my
@ArvinasInc
colleague
@ErikaMVAraujo
and I take a deep dive on oral absorption of PROTAC degraders. We show why assessing rat PK matters, derive clear physchem property limits, and more. Check it out!
This one simple fact is the one thing I wish I could drum into people who don’t do drug discovery for a living. The failure rate and cost of trying are so absurdly high. And all of those failures still have to be paid for.
Clinical development is crazy. So much uncertainty, in the end you can't cheat Science and this is what makes or breaks companies. Ballpark cost for each new drug attempt is at least a couple hundred million $$.
Things are coming back to earth for AI drug discovery - clinical trials are a harsh mistress. fwiw I think there will be future AI successes in the clinic and we can’t pass summary judgment on the back of 3 trials. But a dollop of realistic expectations and humility never hurt.
NEW: The first 3 AI-designed drugs got lots of fanfare when they entered the clinic — not so much when they quietly fell out.
I talked with the AI field about what an 0-for-3 start in human testing means, and whats' next for the field:
For the end of the year, here’s a full roll-up of my
#EarlyCareer
series on demystifying the pharma interview and hiring process (and slightly beyond) - happy reading!
1. Applying for the job / putting a resume or CV together
I’m going to offer some more regular industry
#EarlyCareer
tips on Fridays. Keeping this focused on chemistry in pharma since that’s what I know about.
Let’s start with applying for the job! 🧵 👇 1/
This is industrial med chem in a nutshell. We’re not here to publish papers, we’re here to discover drugs. Publications are on the back burner, typically for IP reasons. Not sure I’d have it any other way either. Discovering drugs is the greater good in the long run.
I’m going to offer some more regular industry
#EarlyCareer
tips on Fridays. Keeping this focused on chemistry in pharma since that’s what I know about.
Let’s start with applying for the job! 🧵 👇 1/
This is a landmark paper and required reading on the proper use of chemical tools for biological validation activities. Especially notable that it’s in a mainstream cancer biology journal. Congrats to the authors, all of whom I know (a rarity).
Ask a medicinal chemist what catalysts they like to have on hand at all times, and you’ll get:
Pd(PPh3)4
Pd2dba3
PdCl2dppf
RuPhos Pd G3
and the like
Simplicity and reliability and broad scope, not necessarily with high yields, are at a huge premium in our line of work.
Unpopular opinion: The distinction between a mechanistically interesting but poorly performing catalyst, versus a more conventional but high-performing catalyst, is lost in the publishing process.
Said another way: complexity is valued over simplicity, which seems backward.
A great perk of my current pharma job, which is fairly common in the sector, is the year-end shutdown from Christmas to New Year’s Day. Kicking my feet up, relaxing with family, expending zero vacation days, recharging without guilt. It’s glorious!
There’s enormous power in converting
“Oh, I didn’t know that” 😞
to
“Cool, I didn’t know that!” 🤩
Build a culture that supports the latter - encourage learning rather than shaming people for not knowing things
You don’t know everything, and neither do I, so lean into it
My PhD research was on the total synthesis of a polyketide natural product. Other than the general skills I learned (lots of reaction chemistry, how to do multi-step organic synthesis), my job today has nothing to do with my PhD.
Don’t get too hung up on the specifics, folks.
So this is pretty wild... the
@disney_lab
has turned dovitinib into a RIBOTAC and in the process repurposed it to target a different *biomolecule* (RNA instead of proteins). In vivo proof of principle too. Lots to unpack.
This is a thoughtful look from
@georg_e_winter
at how what we’ve learned from targeted protein degradation may or may not apply to other proximity induction modalities. Lots of good stuff and well worth a read.
I just read a paper with really poor pharmacology characterization. Y’all know I don’t name and shame on here, but man it’s tempting sometimes. Instead I’ll list some things that could have been done better & also post some fun drinking GIFs that reflect my mood. 🧵 1/