Moneyboy Tweets

[Curtis Cassel]

Stillwaiting to see if anything comes of those lawsuits from all the laid off Twitter employees. My guess is no. Current management may be crazy, but they also have scads of employment attorneys paid to keep things above board.

It\u2019s the start of the new year, and look at that: I\u2019m almost caught up on my monthly Twitter ramble Substack posts \u2014 yay! Here comes the November edition, with usual commentary sprinkled in.

This month\u2019s photo of the month was taken along the eastern ridge of Ragged Mountain in Berlin, CT. This is a challenging hike to a flat-topped summit along the Metacomet Ridge. It\u2019s one of my favorite hikes in the state.

I editorialize extensively when I\u2019m reading; I can\u2019t help it. It rarely makes it into anything I might write in public though. It\u2019s more for organizing my own thoughts. And in fairness, there are also plenty of times when a big ol\u2019 \u201C!!!\u201D will appear in the margin when I find something particularly cool or exciting.

Shortly after the change in ownership of Twitter, the great Mastodon & Post migration began. I was not a fan at the outset, and I\u2019m still not really. The audience is small but the echo chamber is big. I feel like there\u2019s a lot more opportunity to reach folks and build an audience on Twitter. Since I\u2019m a user of \u201Clatest tweets\u201D rather than the algorithmically-driven \u201Ctop tweets\u201D, a lot of the garbage that people are upset about seems to have never made its way onto my screen. There\u2019s a surprisingly good deal of control and curation you can exercise over your Twitter feed, but I sense that a lot of people just go with the defaults and haven\u2019t tried to take greater control. It doesn\u2019t help though that Twitter has done a terrible job of advertising and educating on their customization controls. It helps even less that their default algorithm is dogshit. But I don\u2019t think that\u2019s anything new under the new ownership: the algorithm has always been dogshit.

This theme of life science PhD labor market economics just won\u2019t stop coming up. There are some serious structural issues to address here. The main thing is making sure folks are aware of all of their employment options while they\u2019re still pursuing their PhD, and not locking in on academia as the only/preferred option. I\u2019d much rather view the employment tent as a big one, where we\u2019ll hopefully find gainful employment for most in one sector or another. Chemists have been unusually fortunate over the years to have chronically low unemployment rates.

Hustle culture only seems like a good idea until you get too old to hustle. You\u2019ll wake up one day and realize that some of the best years of your life were filled with work and little else. My biggest regret from when I was an undergraduate student? That I didn\u2019t have a little more fun while it was still easy to do without obligations.

We as a discipline can do more to build a holistic approach to employment. Let\u2019s go out of our way to show people all of the options early and regularly during their educational journey, and trust that they\u2019re smart enough to forge their own paths. Cultivating the notion that academia is the peak career is\u2026 a little cultish. I sense there are some PIs out there who believe they increase their own glory by having a network of glittering academic PI jewels who came out of their labs. But that\u2019s more about the PI than the student, isn\u2019t it? Shouldn\u2019t the right course of action be to celebrate the career successes of all students, regardless of which path they eventually take?

Maybe it was just a slow news month, but it felt like it was time to set down a handful of truths that I\u2019ve come to appreciate about drug discovery over the years. This one is about not getting too caught up in the hype. Executives tend to love the hype because they\u2019re looking for something to fix all the problems this business has. It\u2019s an illusion. Use the tools available, but be aware that they\u2019re just that and nothing more: tools.

The person who\u2019s telling you they\u2019ve found some amazeballs way to shave a year off any part of the drug discovery operation is either delusional or, more likely, riding a giant Dunning-Kruger high because they haven\u2019t been doing it for long enough to know better. What goes around, comes around. This is why I currently have such an aversion to the Silicon Valley invasion of drug discovery. That crowd is used to every problem being reduced to an engineering problem that can be coded into submission, and human biology isn\u2019t that neat and tidy.

A year in the making, this was fun to write with my colleague Erika Araujo. Sadly it\u2019s probably already time for a sequel book. TPD is moving so fast these days that the chapter we submitted in the late fall of 2021 is getting a bit dated. Clinical trials in particular have really taken off across the sector in the last year or so.

This, I think, is the real art of drug discovery: honing your senses to know when it\u2019s right to take calculated risks and say yes. We\u2019re still wrong a lot, of course. But I can\u2019t help but wonder how many might-have-been drugs went to their graves for bullshit organizational and administrative reasons.

Another perennial holiday dinner conversation with opinionated relatives. Between Elizabeth Holmes and Pharma Bro, sometimes it\u2019s a race to the bottom. Laypeople don\u2019t know the difference between these hucksters and the rest of us, and they don\u2019t care either. No matter how much we try to distance ourselves from these kinds of shenanigans, to the general public, we\u2019re largely no better than our worst constituent part. No idea what to do about that either, except for us to be better and call out bad behavior when we see it.

This one is really about proper time management. Drug discovery cycle times are already long, and we can do our part to not make them longer by planning appropriately for syntheses. There are forever hidden needs for additional milligrams (or grams later in the process!) \u2014 so many that it\u2019s easier to just pad the synthesis than to try and rationally predict them all.

Time management again. A simplistic way to frame this calculation: pretend you have a blockbuster drug on your hands with a potential $1 billion in annual sales. Since patent lives are finite and you\u2019ll face competition from generics sooner or later (ever sooner these days), every day of waiting once the patent clock starts running costs you $2.7 million a day in sales. Even if you factor in there\u2019s only a 1 in 20 chance that your drug/project will make it (about right given current clinical failure rates), that\u2019s still $135K a day. The tens of thousands of dollars we\u2019re talking about to run studies in discovery is usually well into \u201Csmall potatoes\u201D territory in comparison. Skimping on stuff in discovery winds up getting done for budgetary reasons, but it\u2019s often penny-wise and pound-foolish.

Beyond my love for True Grit, whether it\u2019s Charles Portis\u2019 novel or either of the movie versions (the Coen Brothers adaptation is far better and more faithful to the book), Rooster Cogburn is our spokesman. If you can\u2019t play a long game, drug discovery is not your game. Or at least not one you\u2019re likely to find much happiness or fulfillment from playing.

Especially true in medicinal chemistry, where so few of us have a hardcore background in medicinal chemistry from our graduate school days. Historically, most medicinal chemists have trained as synthetic organic chemists. Industrial pharma outfits know what they\u2019re getting with new hires: folks who know how to make molecules who are going to have to be trained on just about everything else related to the craft. Sometimes the new hires themselves don\u2019t realize this, however. I\u2019ll also note that I\u2019ve seen some turning of this tide in the last few years, with more people being hired out of academic med chem programs. Even there though, the practice of medicinal chemistry is often very different in industry than it is in academia. For one, they have different goals. Industry is exclusively focused on discovering drugs. Academics may be doing that as part of their program, but it\u2019s rarely all of it. Going along with that difference, there are vastly different scopes in the number of assays (particularly in vivo) and, frankly, dollars that industry has access to.

Thanksgiving is my favorite holiday. An old saying goes that you shouldn\u2019t trust a chemist who can\u2019t cook, so I take my measure of pride in the many years that my wife and I have hosted Thanksgiving at our home. Everything begins with good turkey stock, which I always make from scratch the day before. Stocks and soups almost universally need \u201Cmelding\u201D time and improve with a little aging. An overnight stay in the refrigerator makes a big difference.

Already wrote about this elsewhere, but I think a big chunk of being satisfied with Twitter is how much control you take over your own feed. Ditching Twitter\u2019s algorithm was my #1 quality of life improver.

When I started my Twitter account 6 years ago and began posting about drug discovery and my little corner of the medicinal chemistry universe, I never imagined it\u2019d attract much of a following. And maybe by Twitter standards, 5000 followers isn\u2019t huge. But it\u2019s about 4900 more than I thought would care about what I had to say. So thanks for coming along!

This is really two issues rolled into one. First, the absurd number of significant figures. If someone tells you they\u2019ve calculated an IC50 accurately and precisely to three decimal places, they\u2019re either: a) delusional; b) lying about it; or c) spent their entire PhD program generating replicates to statistically power such an observation. Of course, the standard deviation tells you what the right number of significant figures is, and it\u2019s a hell of a lot less than three decimal places! But the second, bigger issue is the use of arithmetic statistics on geometric data. Anytime you see a mean +/- standard deviation on dose response data, it\u2019s wrong, full stop. Geometric means times/divided by a geometric standard deviation is the only appropriate way, unless the data has been log-transformed. Normal error in a well-behaved biological assay is 3-fold, or half a log. Anyone who isn\u2019t telling you that? Yeah, they don\u2019t know what they\u2019re talking about.

3a8082e126