Below are collected Twitter ramblings for July 2024. Happy reading!
This month’s photo of the month is from the summit of West Rattlesnake Mountain in Holderness, New Hampshire. Squam Lake dominates the view in the distance.
This is needlessly cheeky, and maybe that’s why I love it. Another perennial favorite is the one “quoting” Thomas Jefferson on July 3, 1776 as saying: crap, that’s due tomorrow?
I’m not a regular attendee of any GRC, not because I don’t like the concept, but simply because they fall mostly in the summer months when I’m taking vacations while the kids are out of school. This was my first Drug Metabolism GRC, and it was an amazing experience. Although I’m not a PK/ADME expert by training, I’ve picked up a lot of practical experience with 20+ years on the job as a medicinal chemist. This conference was great because it wasn’t centered in my discipline — but it’s adjacent. There are so many opportunities to learn from experts at meetings like this, and then take what you’ve learned home to apply in your day-to-day work. It’s also why I’ve been a regular attendee of the American Association for Cancer Research (AACR) annual meeting rather than the ACS meetings.
One thing I detected among some of the GRC attendees was a certain disdain for medicinal chemists’ often simplistic understanding of PK/ADME, which we then go on to wield as a blunt instrument. Although I don’t count myself in that category because I’ve worked hard to build up my knowledge in this area — otherwise I suppose I wouldn’t have been invited to speak — the criticism has some foundation in reality. Medicinal chemists need to learn about a large number of adjacent disciplines to do their jobs successfully, including PK/ADME, toxicology, disease biology, assay screening technology, etc. Often we learn just enough to be dangerous, and that’s in turn because we’re teaching ourselves, sometimes the hard way. True mastery of medicinal chemistry as a discipline, though, requires a deeper level of understanding. Not to the level of a trained expert, but enough that you can have a conversation with an expert without them raising their guard that you’re going to go do something foolish.
In equal measure, the thing I think we as medicinal chemists need to feed back to our DMPK experts is that we’re usually in screening mode and any assay they hand us needs to have enough throughput to establish an SAR. I saw some really cool studies presented at the GRC, particularly on measuring drug distribution, that were also finicky and ultra-low throughput. Dangling that kind of data in front of a medicinal chemist is going to lead to them asking you to do the same thing for 10 more compounds. Running a successful drug discovery project requires those minds — and assays — meeting somewhere in the middle.
Just this session alone would’ve been worth the trip. I learned so much from all the speakers, including some newer techniques for measuring free fraction that can be deployed broadly. There was also a great discussion on measuring free fraction in tumors and how to develop good surrogates for that. I was also pleasantly surprised that although some challenges have been mounted to the free drug hypothesis, the consensus in the room seemed to be that there are relatively few exceptions that stand up to close scrutiny.
I’ve made no bones about my disdain for the proliferation of targeted protein degradation conferences assembled by for-profit conference organizers. They try to attract the best and brightest in the field, and then charge an arm and a leg to everyone else to come hear them speak. Who’s really winning here? It’s the conference organizers. Just the registration fees for industry attendees can run $2-3K, and that’s before hotel and meals.
For the speakers, all of whom know the field — and each other — let’s be honest: the delta in the 2-3 months since the last conference is small, and you wind up hearing a lot of the same content over and over again, with minor variations and increments. To me, a GRC is the perfect solution for our field: it democratizes things and returns the power to the scientists who are doing the work. By the scientists, for the scientists. They’re small, intimate meetings that encourage making new connections. They’re also relatively cheap: my GRC cost less than $1000 to attend, inclusive of lodging and all meals. Furthermore, GRCs carry a new data disclosure mandate, which will encourage people to share their latest unpublished stuff in a format where it isn’t going to be broadcast all over town after the meeting.
There’s a definite market for someone to open a real deli that slices fresh meats and cheeses in an airport or train station. Many people — myself included — would happily pay a premium to not have two slices of ancient turkey breast hidden in a day old baguette with soggy lettuce and tomato slices.
I don’t know how else to say this: my job is not business development. Most companies of reasonable size have BD teams to do this work, so please reach out to them instead. It’s part of their job to make sure good opportunities get in front of the right internal people for evaluation. I use LinkedIn for networking with scientists and other industry professionals who mostly share similar interests — not for sales pitches. It’s a great way to reduce the six degrees of separation and can help to generate new job leads if you find yourself in need of one.
I’m also unclear on the value of generically branding yourself as a “skilled pharma professional” — that term is so vague and could mean so many different things that it borders on useless. While it’s probably an attempt to avoid pigeonholing, I read it as evasive. Tell me what you’re good at, and don’t obfuscate it! In the end, if I need to hire an experienced medicinal chemist, there’s going to have to be a track record of being a medicinal chemist in your CV. Yes, having transferable skills is important, especially for entry-level positions. And yes, sometimes companies will bend a little and consider someone with 5-6 years of relevant experience when the job description says 8-10. But no, they’re not going to consider folks who have no relevant experience.
Hint: I’m about three rows directly behind the conference chair John Harrelson.
This mistake is surprisingly common, and it’s something that I always flag during peer reviews. Sadly, the review and editorial process at many journals is slipshod enough that capital ‘H’ Hook effects often make it into print. I regret to report that if you want a proper name of this type, you’re going to have to borrow Hooke’s Law (and an extra ‘e’) from the physicists.
We do seem to want to assign credit to things in this way though; maybe it’s human nature. Perhaps nowhere is this clearer than in the naming of blotting techniques in biology, where we have Southern blots (for DNA), northern blots (for RNA), and western blots (for proteins). (There’s also eastern blots in some circles, but the history and usage there is a bit complicated.) Only the first of these, the o.g. Southern blot, is named after the person who discovered the technique, Edwin Southern. I regret to inform those of you who didn’t know that northern and western blots are simply a cheeky homage to Southern’s name. People who are sticklers for this sort of thing will correctly point out that “western blot” is proper usage and “Western blot” is not, because here western is not a proper noun.
At this point, I’m staying off the “For You” feed on Twitter/X, at least until the election is over here in the US. There are so many trolls across the political spectrum talking nonsense, and they all share some common bad-faith traits: failure to reason with facts and data, engaging in changing the subject and whatabout-ism, and endless ad hominem attacks. It’s not worth a second of time to engage with these folks, because they’re either far-gone partisans who have no intention of changing their minds or they’re influencers who are doing it on purpose for the clicks — or both.
Machine learning models are at their best when they’re making predictions adjacent to the training data, and there are definite uses for those kinds of models in the drug discovery setting. Unfortunately, the problems that need tackling to have the most impact — target selection and clinical trial failure (which are in many ways connected) — are the problems where the available training data, frankly, sucks. If anyone is selling a solution to those problems without a massive proprietary data set to back it up: caveat emptor.
There’s an old joke that academia has infinite time and no money, and industry has infinite money and no time. While not true in the extreme, there’s a kernel of truth to it nonetheless. The pharma sector is competitive and profitability is time-limited by patents, so there’s a big premium on moving fast. I’ve witnessed many teams listing out the project’s problems and issues and brainstorming potential solutions. The best teams recognize early and set aside the “academic” problems — the ones that might lead to a nice publication but aren’t on the critical path to nominating a clinical candidate.
I think some people confused this meme as being adjacent to the Ben Affleck smoking meme, indicating fatigue or weariness. So for those who didn’t watch the Olympics, this is gymnast Stephen Nedoroscik, more affectionately known as Pommel Horse Guy. Nedoroscik only competes in one event — you guessed it, pommel horse. The meme was built on this photo of him getting himself into the zone before his routine, which he proceeded to nail and helped catapult Team USA to a medal in the men’s all-around competition. The meme should thus be taken as someone mentally preparing themselves to go kick some ass on their thing.
I don't often enter the chaos realm that is Twitter anymore, but I love the monthly updates. As someone who loves med. chem. sci. computing that AI post really resonates with me. My first question anytime we're talking with a software company that pitches AI is what's your training set? Because most of the time it's going to be ChEMBL and I haven't seen that set be very representative of modern drug discovery space.