Skip to Content

Lessons For a New Medicinal Chemist

I gave my talk at the Drew University Medicinal Chemistry course, and it got me to thinking about when I was there (1990 or 1991), and my early days in medicinal chemistry in general. There are a lot of things that have to be learned when coming out of a synthetic organic chemistry background, and a few that have to be unlearned. I’ve written about some of these in the past, but I wanted to bring together some specific examples:
1. I had to appreciate just how strange and powerful metabolizing enzymes are. I approached them from the standpoint of an organic chemist, but p450 enzymes can epoxidize benzene, and I don’t know any organic chemists that can do that too well. Ripping open piperazine rings, turning cyclohexanes into cyclohexanols – there are a lot of reactions that are common in metabolic clearance that are not, to put it lightly, part of the repetoire of synthetic organic chemistry.
2. I also had to learn a rough version of the Lipinski rules – basically, that physical properties matter, although the degree to which they matter can vary. You can’t increase molecular weight or lipophilicity forever without paying for it. Small polar molecules are handled fundamentally differently than big greasy ones in vivo. This was part of learning that there are many, many different potential fates for small molecules when dosed into a living animal.
3. Another key realization, which took a while to sink in, was that biological assays had error bars, and that this was true whether or not error bars were presented on the page or the screen. Enzyme assays were a bit fuzzy compared to the numbers I was used to as a chemist, but cell assays were fuzzier. And whole-animal numbers covered an even wider range. I had to understand that this hierarchy was the general rule, and that there was not a lot to be done about it in most cases (except, importantly, to never forget that it was there).
4. As someone mentioned in the comments here the other day, alluding to an old post of mine, I had to learn that although I’d been hearing for years that time was money, that grad school had been a poor preparation for learning how true that was. I was used to making everything that I could rather than buying it, but I had to reverse that thinking completely, since I was being paid to use my head more than my hands. (That didn’t mean that I shouldn’t use my hands, far from it – only that I should use my head first whenever feasible).
5. I also had to figure out how to use my time more efficiently. Another bad grad school habit was the working all hours of the day routine, which tended to make things stretch out. Back then, if I didn’t get that reaction set up in the afternoon, well, I was coming back that evening, so I could do it then. But if I was going to keep more regular working hours, I had to plan things out better to make the best use of my time.
6. There were several big lessons to be learned about where chemistry fit into the whole drug discovery effort. One was that if I made dirty compounds, only dirty results could be expected from them. As mentioned above, even clean submissions gave alarmingly variable results sometimes; what could be expected from compounds with large and variable impurities from prep to prep? One of my jobs was not to make things harder than they already were.
7. A second big lesson, perhaps the biggest, was that chemistry was (and is) a means to an end in drug discovery. The end, of course, is a compound that’s therapeutically useful enough that people are willing to pay money for it. Without one or more of those, you are sunk. It follows, first, that anything that does not bear on the problems of producing them has to be considered secondary – not unimportant, perhaps, but secondary to the biggest issue. Without enough compounds to sell, everything else that might look so pressing will, in fact, go away – as will you.
8. The next corollary is that while synthetic organic chemistry is a very useful way to produce such compounds, it is not necessarily the only way. Biologics are an immediate exception, of course, but there are more subtle ones. One of the trickier lessons a new medicinal chemist has to learn is that the enzymes and receptors, the cells and the rats, none of them are impressed by your chemical skills and your knowledge of the literature. They do not care if the latest compound was made by the most elegant application of the latest synthetic art, or by the nastiest low-yielding grunt reaction. What matters is how good that compound might be as a drug candidate, and the chemistry used to make it usually (and should) get in line behind many more important considerations. “Quickly”, “easily”, and “reproducibly”, in this business, roughly elbow aside the more academic chemical virtues of “complexly”, “unusually”, and “with difficulty”.

33 comments on “Lessons For a New Medicinal Chemist”

  1. WCA says:

    4. As someone mentioned in the comments here the other day, alluding to an old post of mine, I had to learn that although I’d been hearing for years that time was money, that grad school had been a poor preparation for learning how true that was. I was used to making everything that I could rather than buying it, but I had to reverse that thinking completely, since I was being paid to use my head more than my hands. (That didn’t mean that I shouldn’t use my hands, far from it – only that they should use my head first whenever feasible).
    This is one that a lot of folks struggle with. I still have colleagues that worry about how much a compound costs (and bring it up constantly in group meetings), even when you explain to them how much *they* cost the company on a daily basis.
    A couple of grand for 10g of a monomer that will take you the better part of a week or two to make is actually money saved.

  2. Cellbio says:

    Good set of lessons. Learning to be productive while working more ‘professional’ hours than grad students or post-docs is key to productivity in industry. Done right, I think it is even more productive that the long hours of grad school when you factor in quality and not doubling back for do-overs.
    Cell and animal errors bars also key. I grew to so dislike point values for estimates like IC50s that I mandated all IC50 data be present with 95% CI. Without CIs, nobody would know the range of possible true values for a point estimate covers a ten fold range or more and people would make potency ranking calls to select compounds that were not in fact reliable. One way I amuse myself when IC50 curves are shown is to see how much off the line the measured data points lie when there is a beautiful curve (selected in the graphing options for a fixed top, bottom and hill slope). It is also easy to get a sharp inflection when no measured points are near the IC50, only on the top and bottom and widely spaced.

  3. NoDrugsNoJobs says:

    Thanks Derek, these resonate strongly with my own learning curve. I also learned along the way that you have to challenge the folks around you (and accept their challenging you) because we are all amateurs in this, really. I think the difference between many compounds that make it through and those that don’t is advocacy and persistence on the behalf of its “owners”. The path from the flask to the market is beset with many near death experiences and so often, the advocacy and persistence and constant challenging of the conventional wisdom or assumptions was the difference. In my mind I can still envision the story of the lipitor team leader on his bended knee begging management to pass the compound through for further development.

  4. Ellen Clark says:

    These are very useful and helpful points and I will direct any PhD candidates or recent grads, who contact me, to check them out.

  5. CMCguy says:

    Very excellent summary of transition from academic synthetic chemist to a med chemist. One implied lesson is that need to learn to work with others, which often is not part of grad school training, as deal not only fellow chemists to coordinate efforts but more importantly the biologists, comp chemists and many various disciplines can apply or help improve efforts in the lab.
    Although I agree point #8 is relevant view for the med chemists for process chemists need to use a different dictionary than med chem when comes to implementation of “Quickly”, “easily”, and “reproducibly”.

  6. Pete says:

    Assays have error bars but they are also subject to systematic error which can be much more difficult to detect and quantify. One measure of the power of an assay is the weakness of the affinity that it can reliably quantify. Very weak and very strong binding can both be difficult to characterise. A screening assay may sacrifice dynamic range for throughput but that does not make it a ‘bad’ assay.

  7. RespiSci says:

    A difference I see between academia and industry is the idea of owning your data. In academia, there tends to be a desire to be protective of your data and not share, which can be understandable depending on the culture driving the need for publications, and if one has been scooped in the past. Moving to industry, one appreciates that while data belongs to the company, you are expected to have ownership of a project and be accountable yet not be possessive. Making this distinction is challenging for some.

  8. CanChem says:

    re. point 8, the CSO at my first company taught me that in medchem there are only two yields for a reaction: Enough, or Not Enough. If the compound comes out good on a screen, then futz around improving your route. Otherwise, you’re just wasting time.

  9. petros says:

    Cost of chemicals is indeed trivial at the med stage. It’s only an issue later when process development starts.
    It was fascinating to find out how much more cheaply D-glutamic acid could be obtained, in large quantities, than it could from Aldrich et al.
    Also critical is the reliability of suppliers

  10. Imaging guy says:

    Now China has enacted compulsory license law.
    “Exclusive: China amends patent law in fight for cheaper drugs” (reuters)

  11. alig says:

    You should add the lesson: Network and learn skills transferrable outside of mediciinal chemistry because the employment situation is dire.

  12. JasonP says:

    “3. Another key realization, which took a while to sink in, was that biological assays had error bars, and that this was true whether or not error bars were presented on the page or the screen. Enzyme assays were a bit fuzzy compared to the numbers I was used to as a chemist, but cell assays were fuzzier. And whole-animal numbers covered an even wider range. I had to understand that this hierarchy was the general rule, and that there was not a lot to be done about it in most cases (except, importantly, to never forget that it was there).”
    Hah SEE! In an older blog post I mentioned how few chemists have a realistic idea about what biological data is, and here you have admitted it is true. The biggest disconnect between chemists and biologists is chemists live in an absolutist world just like engineers.
    My point was this is why biologists have more longevity in this field, because absolutist thinking is actually easier to wrap your mind around than getting biology to talk to you.

  13. processchemist says:

    @10
    Such a shame… what an unexpected move… all these enthusiastic projections about the growth of patented products in such an axplosively expanding market suddenly turned to garbage…

  14. Josh says:

    Sounds like one damn good talk to me.

  15. CMCguy says:

    #10 Imaging guy I was thinking this was like “Shutting the Barn Door after the Horse is out” but the better way to say may be “Opening the Barn Doors so the Horse can get back in” after of course eating another farmers oats and sexual dalliances with another farmers Mares.

  16. Ted says:

    Hi Derek:
    Good points, all. I started in industry around the same time as you, although my first stint was in a process group. I was dumbfounded when I explored the cost profile of a natural product derivative we had prepared. Within a few steps, virtually the entirety of the compound’s value lie within labor and overhead.
    For kicks, we started substituting the most expensive raw materials you could think of, stoichiometric gold and platinum, highly refined plutonium, etc… it just didn’t matter. After a raw material has spent a few weeks spinning around a reactor train, shoveled to and forth, bagged and unbagged, dissolved and crystallized, etc.. its intrinsic value climbs exponentially.
    It was a great help when I moved to medicinal chemistry. I had no compulsions about scaling intermediates to the hundreds of grams on short notice. Throwing away 80g of unused intermediate is much more cost effective than synthesizing 20g in four batches…
    I think all of the wastefulness, however economically efficient it is, still kills some little part of me every time.
    -t

  17. Ed says:

    #5 – having obtained my Ph.D. in a medical school, I routinely heard how many hours (typically >90) my MD-wannabe colleagues were spending. After a short while, I realized that maybe 50-60 were for school, the other 30+ were in the library complaining about he time spent at school. (The residents actually worked, since they could see patients.) Even now in a mid level pharma company, we spend inordinate amounts of time griping in meetings as the dollars are flying out the window.

  18. Starvinmarvin says:

    …I’d been hearing for years that time was money, that grad school had been a poor preparation for learning how true that was. I was used to making everything that I could rather than buying it, but I had to reverse that thinking completely, since I was being paid to use my head more than my hands. (That didn’t mean that I shouldn’t use my hands, far from it – only that I should use my head first whenever feasible)…
    That doesn’t sound like using your head instead of your hands, now does it? It’s using your wallet instead. While it’s definitely easier to use your wallet instead of your skills, time and your creative mind, in the long run as others have stated very often one would find such an aproach more costly than it looked in its first instance.

  19. Hap says:

    It’s certainly much more intellectually rewarding to build your own plane and fly it, but if I’m in Chicago and I have to be at a meeting tomorrow in New York I don’t think it’s going to work very well. If I use my brain, then I am going to be buying a plane ticket.
    Grant got an awful lot of Union soldiers killed, but he helped win the Civil War because he was willing to use his men (and get them killed) to achieve his country’s ends in battle. Lots of generals before him had been more hesitant to use their men (and the advantage they gave their armies) and thus got nothing for them. You use what you have (thoughtfully) because that’s what it’s there for.
    The labor costs in industry are big, but what could be even bigger (though for most people isn’t) is the opportunity cost of getting a drug out faster. A billion-dollar drug makes nearly $3M a day, and Lipitor at its peak made close to $40M a day. Getting your patent faster, getting to market faster (and getting the right compound there) can be worth a lot of money, amounts which dwarf the benefits of trying to finesse your way to a target.

  20. Jesse says:

    Not sure if cyclohexane to cylohexanol is the best example of a difficult organic synthesis…
    http://en.wikipedia.org/wiki/Cyclohexanol
    “Billions of Kg’s annually”

  21. SelfishAcademicGuy says:

    While on the subject of oddities in this spot-on post, how about RespiSci’s historic first assertion that it’s academics who have difficulty ‘sharing’. Classic.

  22. Chris says:

    But don’t forget the positives
    I still get goose pimples when I think about the day we got the first positive results from the clinical trial.

  23. jackgg says:

    Thanks Derek.

  24. c says:

    Awesome, very grateful for the thought that goes into these comments. Thank you.

  25. Kevin says:

    Generalize it a bit, Derek, and it applies to industrial chemistry in general. For example on 4: We had a summer intern who was working on making a polymerization and when he mentioned needing to make the precursor at a group meeting, got stared at until someone asked “Why not just buy a kilo?”
    You can come up with examples for all of them from the other areas of chemistry. Great post.

  26. Entreprenuer says:

    Derek,
    Great summary, vary important points.
    One more, I think very important point for medicinal chemist is to trust your judgment while designing new analogs. There is definitely something called “Chemist’s instinct”. Lipinski rules, computational chemistry etc is fine, but, is no substitute to real chemist’s understanding of SAR and their intuition.

  27. Chemjobber says:

    Dumb question from a non-medicinal chemist: how do piperazine rings get opened by enzymes? Oxidation to lactam and then hydrolysis?

  28. Chemjobber says:

    OK, so it’s not a lactam, but you know what I mean. /embarrassment

  29. jackgg says:

    Re 27/28, see “BIOTRANSFORMATION OF THE PIPERAZINE RING AND AN EVALUATION OF
    THE EXPERT SYSTEM METEOR” (a poster) at
    http://www.lhasalimited.org/resources/KL2.pdf
    I think it is quite convincing.
    Jackgg

  30. Anonymous BMS Researcher says:

    Similar general principles apply to biologists like me, even if details differ. One of the hardest lessons — which I am still learning after more than a decade in industry — is that the end product is very different! In academia, bosses and granting agencies want publications. So long as we were publishing, bosses and granting agencies were happy. I basically did whatever seemed intellectually interesting and wrote up what I learned by doing it.
    In industry, the end product is a marketed drug. Nearly every day I catch myself wanting to follow up on something that might be of scientific interest but would not be likely to have an impact on our pipeline.
    This does not mean there is no place for my intellectual curiousity: there is. But it means I must channel my curiousity. The impact of my ideas doesn’t always have to be direct, something that has an indirect impact can add value. But if something I do has an indirect impact on our pipeline then I had better have a clear idea of HOW my work contributes to the work of others.
    I am not a chemist, so I will never make a compound. The primary way my work matters to BMS is if, one way or another it helps BMS do a better job of deciding which molecules proceed from one stage of the pipeline to the next stage.
    A second, but also very important way in which my work matters to BMS is when I am asked to do some analysis in order to answer a question from the FDA or one of its counterparts in some other country: everything we do is under the watchful eyes of regulators.
    Of course many other industries are also regulated, but most only have to meet specific standards. Toyota has to prove their new car will comply with safety and environmental rules. Toyota does NOT have to convince any regulator their new car represents a significant improvement in performance over the best-performing cars currently on the market.

  31. simpl re rule 7 on cost containment says:

    20 years ago, when our company was still a chemistry company, the additives division had a rule that only 3 synthesis steps or less were allowed. What is the Pharma rule today?
    The other classic learning story on diy costs came from an engineering department which built a coin slot onto a standard coffee machine. Of course, it came out magnitudes more expensive than the machine itself.

  32. RespiSci says:

    @ 21 Historic assertion? I am currently living it. I will agree that in industry, between companies there is little sharing due to IP concerns. Within an organization however, scientific information is shared. In my return to academia, I have observed a mentality within the university research lab of “I generated these data therefore they are mine-all mine!”. In industry, it is understood that data are not yours. They belong to the company and you are obliged to share this information with your colleagues in the company. I have seen time and effort wasted in academic labs due to a lack of sharing information of methods or findings. If your lab is open, then kudos to you. That is how is should work.
    As for sharing between labs, upon returning to academia I have found the level of paranoia for being scooped by competing labs heightened as compared to the state a decade ago. I have heard PIs instruct their labs not to divulge information to a visiting scientist/guest speaker whereas before I had no recollection of such an instruction. Perhaps this is just my field and I would welcome to hear that such a thing does not exist where you are.

  33. Jon says:

    @ #14,
    It WAS a damn good talk. Derek, even joinded us at the bar afterwards. It was a great night.

Comments are closed.