Comment by danso | Hacker News Reader

Comment by danso | Hacker News Reader

I know the legend of penicillin makes it sound as if revolutionary drugs were discovered by miracle accident, but I'm guessing the less famous drugs, such as tetracyclene, involved a more "routine", non-sexy methodical-hypothesis-testing process, as described in the OP:

> Dr. Conover, however, became intrigued with two naturally occurring antibiotics that, except for two atoms, were chemically identical. To Dr. Conover, the atoms seemed out of place. Each antibiotic had a chlorine or oxygen atom where he expected to find a hydrogen atom. Would swapping in hydrogen improve the potency of the drugs?

> Using a routine chemical procedure, he stripped chlorine from one antibiotic and inserted hydrogen, creating a more stable molecule. He worked with a single assistant. “I didn’t want an audience if we failed,” he told The Plain Dealer of Cleveland in 1992.

> The result was tetracycline, a powerful antibiotic with fewer side effects than the drug from which it was derived — proof, Dr. Conover wrote, that “a superior drug could be made by chemical modification.”

In computer science and computer engineering, students typically learn about, and then how to implement such fundamental breakthroughs as Hamming codes [0] and implementing ALU's in VHDL [1].

What's the equivalent in chemistry/biology/pharma? That is, how do you get into the field/expertise/mindset in which you're creating and testing out drug compositions in a methodical way that Dr. Conover did when he was able to discover and confirm tetracycline in a matter of months.

I realize this already feels like a silly question. Discovering new drugs is more complicated today. And a single science, such as chemistry, isn't enough to create an antibiotic, because you also have to be able to test the result. And perhaps there's a significant amount of chemical engineering process involved to efficiently create the target antibiotic.

[0] http://nifty.stanford.edu/2011/hansen-hamming-codes/

[1] http://www.utdallas.edu/~poras/courses/ee3320/xilinx/upenn/l...

The field of study that most directly relates to what you describe is Medicinal Chemistry (https://en.wikipedia.org/wiki/Medicinal_chemistry)

It's a fascinating discipline. Typically for any given drug, large numbers of candidates are made. Those candidates are trivially different and go through a battery of tests for toxicity and efficacy. While there's a lot of science to it, the amount we know about biology is too primitive for it to be a pure engineering discipline. There's a certain amount of art to it because it's not enough to make the compound. You need to consider that if it works, some day a person is going to have to make it at scale, so it needs to be easy to make too. Finally, biology is surprising and messy so sometimes a gut instinct tells you to go a certain way.

I once was in a talk where one of the medicinal chemists who developed Amoxicillin described his excitement when they discovered the drug they had made had no taste whatsoever. The standard of care oral antibiotic that preceded it for brief period was horribly bitter (I forget the name). Kids hated taking the liquid form as a result. The team realized that with no innate flavor, the liquid form of Amoxicillin could be flavored with something more subtle than the typical grape or cherry flavors that they were accustomed to using to mask the flavor of drugs. He chose bubblegum.

There's a very nice introductory medchem course on edX with Prof Erland Stevens at Davidson College. It has been through a few iterations already, and another round just started this Monday: https://www.edx.org/course/medicinal-chemistry-molecular-bas...

Some members of the audience may find it interesting as this or that facet of the subject is discussed with regularity on this forum - the FDA, antibiotics, intellectual property &c pp.

Discussion of the regulatory and business environment is included, it's after all part of the environment that med. chemists find themselves in.

I used to work at a biotech doing discovery chemistry. How these discoveries are, well, discovered has changed over time. Various techniques come in and out of fashion.

Natural product chemistry (what lead to tetracycline) was very popular the middle of the last century. Scour the earth for unique natural products, test them for activities, then tweak them to optimize. Very common in the mid-1900s up until the 1970's. Some labs still pursue this route. It's great because nature can come up with much more varied molecules than we can in the lab.

Structure-based drug design was next, where you'd start with some structural lead based on basic science and then create a multitude of derivatives, optimizing as you go for binding and pharmacokinetics. Lyrica is a good example. Start with the neurotransmitter GABA and modify it in a effort to make new anti-epileptics. Interestingly Lyrica worked really well, but its mechanism isn't really through the GABA pathway. Very popular from the 1950's through the 1980's, but still in use today.

High throughput screening (HTS) was big a few decades ago. Use robots to create thousands (millions?) of different products based on known chemistry and a scaffold. Test the products for activity (typically binding to a receptor), fish out the lead and then optimize. One big problem is that the chemistry that works well with HTS tends to give you a ton of compounds in a similar chemical space (i.e. you're not creating compounds that are that different from each other). It basically turned out to be a dud (lots of $$, not much to show for it), with a few good hits, but not a lot more success than prior methods. Most biotechs stopped doing HTS in the 1990s.

Fragment-based drug design has come into vogue although I left the lab before that happened. My basic understanding is that instead of looking for molecules, you look for smaller fragment binding to known targets, then when you find a handful, you link them all together to get your lead. Some exciting results so far, but it's early.

> What's the equivalent in chemistry/biology/pharma? That is, how do you get into the field/expertise/mindset in which you're creating and testing out drug compositions in a methodical way that Dr. Conover did when he was able to discover and confirm tetracycline in a matter of months.

I'm not a chemist, but several times I've walked up to a organic/medicinal chemist and been told something like "putting a fluorine there (pointing at a chemical structure) should make it bind better. I can do that for you in a week or two."

There's apparently whole bodies of work [0], especially in medicinal chemistry, where you methodically put a fluorine at different positions around a ring (or something similar) and you test each new compound. Sometimes one will have some property that's a few orders of magnitude better than the rest, and you get a "miracle" drug.

[0] http://pubs.acs.org/doi/abs/10.1021/jm800219f

Fluorinations are very popular as fluorine is not that much bigger than hydrogen, so you don't tend to cause big confirmational changes with the swap. It's also very stable, so it's doesn't screw up your pharmacokinetics. And finally, fluorine is super polar, so a simple swap can dramatically change the drug's activity.

As someone else said, drug design is still a lot of hand waving and hit and miss. Sometimes you can make the smallest changes to a molecule and have a huge impact. The fentanyls are a good example. Swap a hydrogen for methyl and you can increase potency by 100x.

While I can't answer your question about biology/pharma, I will say that depending on how you look at it - Conover's vindication of routine chemistry had some of the same elements of sexy as Fleming's discovery of Penicillin. That is, both discoveries were "hidden in plain sight" and owe their existence to the different lab work styles (and thus points of view) of their discoverers.

Reply on news.ycombinator.com