Skip to main content

Chemical News

Robotic Flow Synthesis: The Latest Version

Here’s another step along the way to automated synthesis, in a new paper from MIT. The eventual hope is to unite the software and the hardware in this area, both of which are developing these days, and come up with a system that can produce new compounds with a minimum of human intervention. Let’s stipulate up front that we’re not there yet; this paper is both a very interesting look at the state of the art and a reminder that the state of the art isn’t up to that goal yet.

The software end of things involves (in the ideal case) being able to come up with plausible synthetic routes to the desired molecules, with “plausible” being not only in the abstract but fitted to the abilities of the hardware synthesizer itself. And since that synthesizer is very likely going to partake of a lot of flow chemistry, you’ll have a lot of thinking to do about concentrations, flow rates, non-clogging conditions, and so on. I should mention that an even more ideal system would be able to come up with its own ideas about what to synthesize, but doing that from a standing start is even further off. I think what we’ll see before that (and people have already been working on this as well) is a system that can suggest reasonable rounds of analogs given the assay data from a previous round of simple analogs, and then can turn its attention to how to synthesize them.

In this case, the flow of operations looks like this: (1) Select a synthetic target, (2) search the literature for the compound, (3) retrosynthetic analysis, (4) select reaction conditions, (5) estimate feasibility, (6) formulate a recipe for the hardware to follow, (7) configure the platform for that recipe, (8) test run of the process, (9) scaleup of the synthesis in flow, and there’s your product. But as the paper notes, there are still several of these stages that need human input, as you can certainly imagine if you’ve ever done organic synthesis, done any flow chemistry, or worked with automation of any kind at all. One of the good things about this work, actually, is the job it does highlighting just the areas that turned out to need the most human help as they got this system into shape.

The MIT group has been working on their own retrosynthesis software (ASKCOS), and they give some details of it here. The system was trained on millions of reactions abstracted from both Reaxys and the USPTO database, so it’s seen plenty of organic chemistry. For example, out of the 12.5 million or so single-step reactions to be found in Reaxys, the system is set to pay attention only to those with ten or more examples, which knocks things down to about 160,000 “rules” for valid single-step transformations. They then trained a neural network to try to predict which of these rules would be most applicable to a given new target structure – this step was put in to decrease the computational load in the next step and also to try to increase that step’s success rate.

Each proposed retrosynthesis step first gets a binary filter applied to it: are there any conditions that the program knows about that could generate the desired product from the stated reactants? Getting rid of stuff at that stage saves a lot of pointless calculation. Then if the answer is “yes”, the program turns its attention to the hits, and the proposed reaction sequences are evaluated by a more computationally intensive foreward-prediction model, which is trained up on the most synthetically plausible conditions for each transformation. If that result matches up, the route is considered believable.

The hardware is a later version of the system I wrote about here, a deliberately modular plug-and-play setup. This leaves the individual modules themselves open to upgrading on their own without affecting the rest of the system, and cuts down on the number of valves and connections in any given overall plan – and as anyone who’s set up an HPLC, LC/MS, or flow chemistry apparatus will be able to tell you, mo’ connections = mo’ problems. Every new fitting is an opportunity for something to fail later on. There is, as before, a manipulator arm in the middle of the thing that can reach over and plug the individual modules into a common rack to assemble the sequence needed for the synthetic scheme (heated loop flows into phase separator flows into solid-supported reagent bed flows into. . .) There is a lot of engineering involved in getting this to work, a forest of little details that have to be addressed. Just to pick one example, the tubing involved is all tensioned via spring-loaded reels to cut down on looping and tangling – you can easily imagine what might happen otherwise as the robot arm merrily assembles something that look like a colander full of angel hair pasta when you walk around and look at it from behind. The system generates a “chemical recipe file” for a given synthetic path; this CRF has the mapping of the physical reaction setup and the operations needed to execute the synthesis. This includes locations of stock solutions, assembly of the modules, solvents, flow rates, temperatures, and all the rest of it.

The examples given in the paper are part of the “drugs on demand” work that the Jamison lab has been doing for several years now (which I wrote about here). I remain somewhat skeptical of the stated overall DARPA goal of this work, as that post shows, but it’s an excellent proving ground for automated synthesis (which to be fair, is also one of the goals). Links below added by me:

We therefore chose a suite of 15 medicinally relevant small molecules, which ultimately required eight particular retrosynthetic routes and nine specific process configurations. Although literature precedents exist for all 15 targets, the synthesis-planning program is prevented from merely recalling any synthetic route from memory as exact matches; all pathways are required to be discovered de novo through abstracted transformation rules and learned patterns of chemical reactivity. . .In order of increasing complexity, we investigated the synthesis of aspirin and racemic secnidazole run back to back; lidocaine and diazepam run back to back to use a common feedstock; and (S)-warfarin and safinamide to demonstrate the planning program’s stereochemical awareness. . .

They also include (in the SI) a synthesis for benzfibrate that had to be abandoned due to poor flow chemistry performance, which is a detail that I very much appreciate. The paper also includes two small library syntheses around ACE inhibitor and COX-2 inhibitor scaffolds. This is a good time, though, to go back over that list of steps above and note which ones needed (in one case or another) human evaluation. The answer is, most of them. Step one, selecting the target, is of course a completely human operation. Step two, searching the literature, is done automatically. Steps 3, 4, and 5 (retrosynthesis, selection of conditions, and evaluation of feasibility) are a mixture. Human input is available (and desirable) for each of them – users can and should set various thresholds and biases depending on how the runs are going. But I invite my fellow humans not to get too cocky, considering that all of these steps were until very recently considered our exclusive domain. The same goes for step 6 (formulate a CRF) and step 7 (configure the apparatus). The software does most of this itself, but human “reality checks” are needed here as well. As for the other physical steps 8 and 9, actually running the chemistry, the machine obviously does this, but I strongly suspect that there are people standing around watching it while it does, at least at first.

The details of each synthesis are quite interesting, as the software picks out certain reactions and conditions. Even simple steps have a lot of decisions: making aspirin is dead easy, but do you do the acetylation with acetyl chloride or with acetic anhydride? Any added acid catalyst? What solvent? What concentration? What temperature? If you run it neat, you have an increased risk of clogging, for example. Different acidic conditions might be more or less compatible with the downstream apparatus. Some solvents that drive the reaction more quickly to completion could be harder to remove at the end. Every single step in a synthesis, as practitioners will appreciate, involves a list of such decisions. Watching the software deal with them is startlingly like the experience of teaching a teenager to drive: you realize how many small details you take for granted that you have to call attention to. And the flow chemistry aspect introduces new complexities at the same time that it enables the whole idea to work at all – for example, you’re probably going to want to use a soluble liquid base like triethylamine or DBU rather than potassium carbonate, which will have to come in as an aqueous solution in flow, and would generate gas bubbles even if you did. Which base, though, will produce a hydrochloride that’s less likely to clog the system?

That brings up an important point that the paper highlights: we actually don’t know enough chemistry to predict how these things will work, even for rather simple reactions. There is a lot of empirical experimentation every time you set up such a system, for reasons like this:

Approximate conditions for batch synthesis can be generated based on the literature, as we have done in this study, but their direct implementation in flow is challenging. The desire for process intensification (e.g., to decrease reaction times), the need to mitigate solids formation to avoid clogging, and the importance of telescoping multiple unit operations requires deviation from batch conditions and a level of confidence of predictions that flow chemistry has not yet achieved. Computational prediction of solubilities to within even a factor of 2 in nonaqueous solvents and at nonambient temperatures remains elusive. Predicting suitable purification procedures is a general challenge, not just for flow chemistry, particularly when using nonchromatographic methods.

And very much so on! No, these are not solved problems, and you should hold on to your wallet in the presence of anyone who tries to tell you that they are. But at the same time, there is no reason for them not to be solvable. We just need more information and to get better at what we’re doing. Over time, as the paper notes, we will assemble (we already are) a great deal more knowledge about flow chemistry and reaction prediction, and systems such as these will gradually become more and more capable. You can read a paper like this two ways: you can look at the limitations and what remains to be done, or you can look at what’s already been accomplished and how much of that you might have once thought was restricted to human effort. My advice? Don’t neglect either perspective, because they’re both valid.



19 comments on “Robotic Flow Synthesis: The Latest Version”

  1. Barry says:

    In 2019, the single niche that cries out for these “drugs on demand” is radioisotope synthesis. These are the compounds that must be synthesized/purified/delivered again and again because they decay with time (some of them with half-lives of minutes)
    But some will consider that a colony of humans off Earth might rely on the descendant of such a system for all drugs.

    1. anon says:

      Spot on! As someone working in this area (Radio-chemistry) with a background in Medicinal chemistry, you took word out of me! Thanks.

    2. zero says:

      An offworld colony is far more likely to have a chemistry staff (medicinal and otherwise) to handle synthesis instead of investing in complex, expensive and single-purpose hardware like a synthesis machine. The same staff might spend most of their time on industrial process chemistry, so diverting them to make up batches of medication once in a while is worth doing.

      1. loupgarous says:

        Machines are comparatively cheap and don’t mass more than humans when you count life support and food. What’ll happen is most synthesis will be farmed out to the machines, with a few human chemists present to perform sanity checks on what’s being made. But the actual donkey work of synthesis, industrial or medicinal, will mostly be done by machines with humans analyzing what’s actually made for safety and efficacy.

        By the time we actually colonize Mars and deep space (say, Enceladus or other gas giant moons), it’ll be cheaper to send three or four complete robotic synthesis equipment suites, along with analytical gear for humans to independently confirm their output and controls for humans to correct robotic errors, just to make sure the colony has life support in every imaginable way (food, medicine, industrial chemicals to support the colony’s operations and keep it running). We’ll still need a few chemists to supervise the robots, but not to make the actual compounds.

        1. AlloG says:

          Why cant you just put da drugs into vending machines? I saw dis at Schipol last week.

          By da time you send da chemicals, da machines and TLC plates you could just send one vending machine.

          1. ReOrgChemist says:

            Why simple if you can invent a complicated system?

          2. loupgarous says:

            Because a catalog of robotic syntheses on digital media doesn’t mass as much as very possible medication an off-world colony (or a Third World country on Earth) might need in the quantities needed to treat victims of an outbreak of infectious disease. That catalog can also be updated electronically.

            Logistics is rough for an off-world colony – that hypothetical vending machine full of all the useful medications in useful quantities not just for treating outbreaks of rare disease, but everyday episodic illnesses in the colony population will at some point in the future – which is when we’re talking about colonizing deep space – outweigh the robots and their feedstocks.

            As robotic syntheses become cheaper and capable of creating more useful medications, it’s conceivable that in the future the economic lines will cross where it’s cheaper and better for Third World country medical staffs to have a machine that makes the medicines on their version of the WHO Model List of Essential Medicines than to have to stock medications they only need now and then (or have them delivered when they’re needed to save lives).

            As formularies expand to include new drugs, robotic synthesis will make more sense than storing every potentially useful drug in that vending machine.

            Not to mention those drugs may not keep as well on the pharmacy shelf as we’d like them to.

          3. Anonymous says:

            This is a reply to loupgarous: “Because a catalog of robotic syntheses on digital media …” but there is no “Reply” button on that post. I’ll start with the negatives and point out that starting materials for automated drug synthesis also have expiration dates. If you need to synthesize more penicillin you’d better make sure that your CDI or other coupling reagents haven’t expired. In an Earth lab, you toss it in the waste can and buy a fresh bottle from SIAL. Likewise for every other starting material in the recipes.

            Acute vs Chronic: So a colonist is diagnosed with a malady and needs treatment. If it’s a mild headache, “I’ll whip up some pain meds and have them for you in 48-72 hours. Bite the bullet until then,” might suffice. But if it’s an infection or other acute problem, “I’d whip up a really good antibiotic for that but you’ll be dead before the machine finishes so, I figure, why waste good chemicals?” will come as a disappointment.

            Purification. Years ago, it took more than 100 L of water to deliver 1 L of Coca-Cola to a consumer. I think it is now down to 20-25 L per L of final product. Still incredibly wasteful. It’s easy for a non-scientist to visualize that. What would you rather ship to your space colony: one pallet of finished Coca-Cola bottles or 20x pallets of water, plus another unknown number of pallets of syrup or syrup ingredients, a pallet of empty bottles, mixing and bottling equipment, …. The same idea applies to synthesis, whether it’s via solution phase or solid phase. Solid phase synthesis can be even MORE wasteful: recipes often call for larger excesses of reagents because it’s so easy to wash them away with multiple high-ratio solvent washes before moving on to the next step.

            From recent In The Pipelines, we know that final purity checks, QC, and validation is another important issue (e.g., “Tales From FDA Site Inspections, 9 AUGUST, 2019). “Hey loup, here’s that drug you need. It looks pretty clean by TLC … just a couple of extra spots, but not too bad.” In addition to the Synthesis Machine, your space colony lab will want to have numerous RELEVANT instruments to QC each intermediate and final drug. (In those earlier threads, I posted about choosing the RIGHT assays and running them correctly. There is not just ONE universal, simple test for universal drug purity.)

            Formulation. You also have to keep fresh batches of binders, excipients, and so on. And all sorts of sterilization equipment. Different meds are sterilized differently. At least if your GMP starch gets contaminated, you can send it down to the space colony kitchen for them to make a batch of cookies or pizza crust.

            Waste. The waste from a finished drug-store drug is a small plastic bottle or small paper, plastic, metal foil, etc. pill packaging. The TOTAL waste from de novo synthesis of a few tablets will be HUGE. A sensible space colony might be able to recycle some of that waste. Non-recyclable contaminated waste will have to go somewhere, preferably NIMSBY – Not In My Space-colony’s Back Yard.

            If my space colony has the technology to maintain and purify dozens of starting materials, it can also maintain and purify (if necessary) finished products.

            I have already commented that real-world analytical instrument maintenance can be far more difficult than keeping one storage fridge in good working condition.

            If you are going to go with the WHO Essential Medicines list, you can probably also get an estimate of what amounts to keep on hand for a given space colony population for a given period of time. You probably won’t need a lot of malaria meds. You probably will want to bring extra GI meds in case people come down with a need for Colonists’ Colonics for Montezuma’s Revenge or the Spacedoor Trots. (We also have real guidance from 50+ years of space missions, arctic missions, long term submarine missions, etc..)

            But a good space colony will have a chem lab and one or more chemists just in case there’s a need to whip up something special, not in the medical kit.

            I guess what I’m saying is that I think that pre-fab essentials is more efficient than trying to make everything “fresh” BUT that having chem capability will make it possible to augment the First Aid kit, as needed.

            (A popular magazine cover linked in my handle: “You promised me Mars colonies. Instead, I got Facebook.” – Buzz Aldrin, Technology Review, Nov 2012.)

        2. Long Ago and Far Away says:

          As long as I get to watch watch C-beams glitter in the dark near the Tannhäuser Gate, I don’t care who or what synthesizes off-world medicines.

      2. Barry says:

        “single purpose”??
        perhaps you and I are discussing different articles.

      3. Anonymous says:

        Thanks for bringing some real world experience to the off-world. We had service contracts on most of the big expensive equipment. There were field techs from many companies always in the area. Back then, the top of the line mass spec had a lot of down time due to routine maintenance (cleaning the source, replacing seals and worn parts, hi vac maintenance, calibration, etc.). A senior member of the department was complaining to the field tech in the instrument center about the 30-40% down time and the field tech replied, “That’s excellent. Normal down time is 50%.” I have done a LOT of instrument maintenance in my labs and it is a PITA. I’ve done lots of plumbing repair on LCs, replaced defective circuit boards on numerous instruments (after waiting for FedEx delivery of replacements), etc..

        More recently, with state of the art instruments and multiple NMRs with auto-samplers in the instrument centers, they were still hardly ever all working at the same time. Some companies had full time field techs in their own offices on campus!

        These fancy machines are great, but they have not yet reached the level of science fiction perfection.
        Bowman: “You know of course though he’s right about the 9000 series having a perfect operational record. They do.”
        Poole: “Unfortunately that sounds a little like famous last words.”
        … HAL 9000: “I’m sorry, Dave.”

    3. electrochemist says:

      The most realistic scenario for automated off-world synthesis of drugs would be for ASOs. The off-world station would only need to maintain an inventory for a limited number of amidites, which could be assembled on-demand in the required order for any one of a number of drugs. This will require discovery and development of a much greater number of ASO drugs than exist currently, however.

  2. KevinG says:

    I would love to see what happens when you feed it a target that is well-precedented in the literature using ‘classical era’ organic synthesis. Does the robot learn from the experience of getting inexorably constipated from making an amide with DCC? Does it get stuck in an endless scale-up and purification loop when it undertakes a Skraup reaction? Would it be able to optimize a McMurray coupling?

    1. loupgarous says:

      Another excellent reason we’ll always need humans in the loop – because difficult syntheses are done because humans aren’t limited to sequential thinking, they also think laterally (to non-obvious routes in the case of chemical synthesis) as well. Several papers Derek has shown us in this blog illustrate just this happening, molecules not performing in vivo or even in vitro as a reasonable chemist might expect from previous work in the field. Would AI make those same compounds for the same purposes when asked to come up with some good solution to a problem?

  3. This work is already my highlight paper of 2019, this will be hard to top!

    Derek, maybe it is worth pointing out that the synthesis planning software component employed in this work builds on previous work that you already wrote about in 2018 – which this work here significantly extends with more advanced reaction prediction capabilities, and condition prediction. And most importantly, it is finally providing an impressive array of actual experiments that suggest that machine learning in organic chemistry is here to stay.

  4. John Wayne says:

    With current propulsion technology, the most important consideration for bringing anything with you is mass. Bringing the actual medicines you might need weighs so much less than any lab equipment plus consumables, I don’t see any sort of synthesizer being used in the space program in the short or medium term.

    Yeah, warp drives and nanoassemblers could change everything – they always can.

    1. AlloG says:

      Yo Duke! You da man! As I always say to my kick “Those Doggies wont get to Market by themselves, Pilgrim!”

  5. Nameless says:

    You can mostly resolve the problem of flow chemistry and workup by using a solid supported Produkt and attach each group succesivly. I don’t know how well it works in non-peptide chemistry but is simpler than automizing a flow cell to work without human intervention.

    1. Anonymous says:

      Nameless: [use] “a solid supported Produkt and attach each group succesivly. I don’t know how well it works in non-peptide chemistry …” I tend to give credit to Ellman, J. Am. Chem. Soc. 1992, 114, 10997-10998, for the first solid-phase non-peptide, non-oligo, small molecule synthesis. Any other candidates? That paper helped to launch 20+ years of combi chem, much of it of dubious quality and value. Go to Amazon and search book titles for “solid phase organic synthesis.” There are many.

Comments are closed.