Computational planning of the synthesis of complex natural products

Engines of discovery: Computers in advanced synthesis planning and identification of drug candidates

“…Synthia™) designed syntheses of complex natural products ( Figure 1) with precision and elegance the world's leading chemists judged to be indicative of human, expert-level planning. 11 The first part of this article aims to narrate how this was achieved and why the problem turned out to be so difficult to tackle, taking us some 20 years of concerted effort. [12][13][14][15][16][17][18][19] The second part goes further and, inspired by Chematica's success in retrosynthesis, outlines other areas of synthetic chemistry in which computer-driven approaches -in particular forward-synthesis combined with property prediction -can make a profound and lasting impact: In the Origins of Life, in green chemistry, in the generation of molecular diversity, or in the prediction of synthesizable drug candidates.…”

“…In chemistry, the "game" of retrosynthesis is how to choose chemically plausible pathways from the giant network of synthetic options. "Giant" here is not an exaggeration -with 100,000 rules, every retron can produce, in one step, on the order of 100 synthons, 11,15 each of these synthons can then produce ~100 progenies, and so on, until commercially available starting materials are reached. In n steps, this branching translates into 100 n possible routes one can trace on the network.…”

“…We improved the scoring functions, and implemented search algorithms that now used multiple search strategies simultaneously -for instance, one scoring function preferring diversity of approaches ("searching wide"), one putting premium on finishing the routes as rapidly as possible ("searching deep"), and the two exchanging and learning from each other's results. Unfortunately, even with these and other improvements reviewed in 11 , the critics seemed to had been right -the program was not robustly identifying routes to complex targets.…”

“…The improvement was immediately manifest in the program's ability to plan syntheses to complex natural products 11 as illustrated by the synthesis of Methyl Monate C in Figure 1a or the synthesis of Aplysin in Figure 1b, both of which are hardly discernible from routes that a human expert might design. In fact, in a recent paper on the topic, 11 we assembled a collection of 20 Chematica-planned and 20-literature published (in journals like J. Org. Chem., Org.…”

Synthetic biology 2020–2030: six commercially-available products that are changing our world

“…Enzymes can be incorporated into retrosynthetic routes by mining them from databases using DNA synthesis and screening and controlling their specificity using evolutionary and computational methods [25][26][27][28] . Balancing the constraints of chemical-biological routes "by hand" will be nearimpossible, thus requiring design software similar to that which has recently revolutionized chemical retrosynthesis 29,30 . Efficiently manufacturing these compounds will require innovations in determining when steps should be performed in a living chasses versus a cell-free system and innovative modular reactor designs 12,30,31 .…”