Update: the author of this paper has appeared in the comments here (and elsewhere) saying that he's withdrawing the paper. These are apparently reviewer's comments on it, although I have no way of verifying that. Many of them don't sound like the comments I might have expected. There's more here as well.
Here we have one of the oddest papers to appear in Drug Discovery Today, which is saying something. The journal has always ranged wider than some of the others in this space, but this is the furthest afield I've seen to date. The title is "DrugPrinter: print any drug instantly", and I don't think I can do better than letting the abstract speak for itself:
In drug discovery, de novo potent leads need to be synthesized for bioassay experiments in a very short time. Here, a protocol using DrugPrinter to print out any compound in just one step is proposed. The de novo compound could be designed by cloud computing big data. The computing systems could then search the optimal synthesis condition for each bond–bond interaction from databases. The compound would then be fabricated by many tiny reactors in one step. This type of fast, precise, without byproduct, reagent-sparing, environmentally friendly, small-volume, large-variety, nanofabrication technique will totally subvert the current view on the manufactured object and lead to a huge revolution in pharmaceutical companies in the very near future.
Now, you may well read that and ask yourself "What is this DrugPrinter, and how can I get one?" But note how it's all written in the conditional - lots of woulds and coulds, which should more properly be mights and maybes. Or maybe nots. The whole thing is a fantasy of atomic-level nanotechnology, which I, too, hope may be possible at some point. But to read about the DrugPrinter, you'd think that someone's ready to start prototyping. But no one is, believe me. This paper "tells" you all the "steps" that you would need to "print" a molecule, but it leaves out all the details and all the hard parts:
Thus, if DrugPrinter can one day become a reality it will be a huge step forward in drug discovery. The operator needs only to sit down in front of a computer and draw the structure of compound, which is then inputted into the computer, and the system will automatically search by cloud computing for suitable reaction conditions between bond and bond. . .
That actually captures the tone of this paper pretty well - it exists on a slightly different plane of reality, and what it's doing in Drug Discovery Today is a real mystery, because there's not much "Today" in it, for one thing. But there's something else about it, too - try this part out and see what you think:
Thus, this novel protocol only needs one step instead of the five-to-ten steps of the current synthesis process. In actual fact, it is even better than click chemistry, with lower costs and with better precision of synthesis. A world-leading group led by Lee Cronin has made advances with the technology named ‘Chemputer’. However, it is different to our concept. We specifically address the detail of how to pick up each atom and react. We also disagree that it is possible for anyone to simply download the software (app) from the internet and use it to print one's own drug. It is not feasible and should be illegal in the future.
Some of this, naturally, can be explained by non-native English usage, although the editorial staff at DDT really should have cleaned that up a bit. But there's an underlying strain of grandiose oddness about the whole manuscript. It makes for an interesting reading experience, for sure.
The paper proposes a molding process to fit the shape of the desired target molecule, which is not prima facie a crazy idea at all (templated synthesis). But remember, we're down on the atomic scale here. The only thing to build the mold out of is more atoms, at the same scale as the material filling the mold, and that's a lot harder than any macroscale molding process that you can make analogies to. The MIP (molecularly imprinted polymer) idea is the closest real-world attempt at this sort of thing, but it's been around for quite a while now without providing a quick route into molecular assembly. There is no quick route into molecular assembly, and you’re certainly not going to get one from stuff like this:
Benzene has six carbon atoms joined in a ring, with one hydrogen atom attached to each carbon atom. It can be divided into six reactors for three atoms: C, H and C (Fig. 3). After inputting the chemical structure of benzene, the system will search for the best synthesis condition for each bond. The best optimal condition will be elucidated by computer and controlled by a machine with optical tweezers to pick up the reactant and the atoms of carbon and hydrogen. The carbon atom will be picked up by optical tweezers in the right position in these tiny reactors (just like a color laser printer). DrugPrinter technology will work just like a color laser printer but instead of a four-color (red, yellow, blue and black) printer toner cartridge there will be various atoms.
Right. The computer will search for the best reaction conditions for building up benzene by individual carbon atoms? There are no best conditions for that. You can make benzene from acetylene, if you’re so minded, but you need metal catalysts (Reppe chemistry). And how are these “conditions” to work inside some sort of benzene-shaped mold? How are the intermediates (propene? butadiene?) to be held in there while another carbon atom comes in? Making benzene in this manner would be Nobel-level stuff, and this paper’s just getting warmed up:
. . .The chamber for the storage of elements is divided into three parts based on the character of each atom according to the periodic table of elements. Roughly, there are three categories: nonmetals, metals and transition metals. Of course, most drugs are organic compounds, thus it is reasonable to expect that carbon (C), hydrogen (H) and oxygen (O) will be the major consumables (just as the black toner cartridge always runs out before the other three colors in a printer). . .
I don’t know what the author’s background is, but honestly, you get the impression that it doesn’t include much organic chemistry. The whole paper is written about a world where you take individual atoms from these reservoirs and fly them down small channels “with lasers or plasma” to be caught by optical tweezers and put into the right position. Apparently, things are just going to snap together like so many molecular model pieces once that happens. Reaction mechanisms, thermodynamics, reactivity and selectivity make no appearance at all that I can see. What does make an appearance is stuff like this:
Big data is applied suddenly in any field. For DrugPrinter, we allow the user to upload their desired compound by a webserver. A cloud computing system and fast dealing and optimal of the chemical reaction must be searched immediately. All the bond–bond reactions will be collected in an intelligent system by cloud computing. Because we built a world-first intelligent cloud computing drug screening system called iScreen (http://iscreen.cmu.edu.tw/) and an integrated webserver (http://ismart.cmu.edu.tw/) including the world’s largest traditional Chinese medicine (TCM) database (http://tcm.cmu.edu.tw/), this has enabled our technology. . .
I’m not trying to be unkind here, but some of this reads rather like the spam comments that pile up on this blog and others. “The buzzword will be made by high-tech buzzword by use of buzzword systems”. None of this is real, and as speculation it’s not too exciting, either. Eric Drexler is far more interesting reading – you can certainly find many things to argue about with him (as Richard Smalley famously did), but he’s thought about these problems in a useful way, as have many others. Drexler’s name, by the way, appears nowhere in this current paper, although the whole thing reads like a smudged tenth-generation photocopy of his work from the 1980s.
And that brings up an editorial question: who reviewed this? How did the staff at Drug Discovery Today find this worth publishing in its current form? I have no problem with them running papers about speculative nanotech chemical synthesis, I should add. I like that stuff; I like reading about it. But I don’t like reading hand-waving hoo-hah illustrated with videos of traditional egg-cake molding machines (I kid you not). As published, I found this paper to be an irritating, head-shaking, eye-rolling waste of time, and I would gladly have said so in a referee report. I see that Chemjobber is baffled as well. Who wouldn’t be?