Chemists in the US are trying to create what they describe as an ‘internet for chemistry’, a network linking millions of chemical substances – and the reactions that link them – which could be used to discover new and better ways to synthesise molecules.
The Chematica project has been ongoing since 2005 but hit a milestone over the summer when a series of papers on the project – published in the journal Angewandte Chemie (August 6 edition) – provided preliminary evidence that the platform could quickly find alternative, simpler synthesis routes for some common organic molecules.
The team behind the project is run by Bartosz Grzybowski of Northwestern University in the US, who believes an early application for the project could be to find cheaper and less environmentally-damaging ways to synthesise pharmaceutical ingredients.
Chematica is an attempt to link chemical compounds and the reactions between them into one database, creating “not only a new repository of chemical methods, but an entirely new knowledge platform where each chemical reaction ever performed and each compound ever made would give rise to a collective ‘chemical brain’,” says Grzybowski.
The project arose out of a realisation that chemistry – unlike many other academic fields including scientific computing – had somehow failed to tap into the potential of networks that explore the inter-relationships between things. It is based on algorithms such as those used to power search engines like Google that have been adapted for chemical research.
Immortal chemist
Chematica attempts to look at chemistry as a network of organic molecules linked by the reactions between them, and includes every single reaction reported in the literature over the last 250 years, which Grzybowski describes as a “monstrous construction”.
In effect the network operates as an ‘immortal chemist’ – an expert on the chemical research team who never retires or dies with the resulting loss of his or her accumulated knowledge and experience.
Chematica’s practical value lies in the fact that it enables identification of the most useful chemicals that can be used in a reaction and predicts the efficiencies of new chemical transformations, as well as the properties of the most likely products. It also includes data on the prices of chemical precursors to provide a view of the cost of a reaction.
In essence, the platform is a “global view on chemistry that is not based on the knowledge or expertise of a single chemist but of every chemist who ever lived”, says Grzybowski.
“By analysing [the database] with the tools of network theory and statistical physics, we discovered the laws that govern all synthetic transformations carried out to date or to be carried out in the future,” he says.
All told, Chematica contains data on around seven million compounds connected by a similar number of reaction pathways and some 86,000 chemical rules – many times more than could be learnt by even the most accomplished organic chemist, according to its developers.
Exploring synthetic pathways
Chematica’s algorithms can be used to explore all possible synthetic pathways – not just those that spring most readily to the chemist’s mind – and provide various possible options based on predefined criteria, such as the most environmentally-friendly route or the one that makes use of the cheapest precursors.
The three papers published in August provide preliminary evidence that these claims may be borne out.
Paper one: The first paper describes how the software has been used to design more economical syntheses of commercial chemical products.
Grzybowski explains that in the classical environment a chemist will design a synthetic pathway for a molecule by charting a route back through intermediate molecules a step at a time until a recipe of precursors is identified. At each stage there may be dozens of alternative pathways but chemists will often overlook alternatives in favour of tried and tested routes.
Synthesis can be optimised with various constraints, such as avoiding reactions involving environmentally damaging or hazardous compounds, and the paper found that more than 50 reactions could be reduced in cost, in some cases by up to 45 per cent.
Paper two: In the second paper, Grzybowski’s team used Chematica to try to find ‘one-pot’ alternatives to multi-step syntheses for two groups of chemicals – quinolines and thiophenes – in a bid to reduce separation and purification steps that eat into yields and make syntheses lengthier and more laborious.
They discovered a number of alternative synthetic pathways that were shown in the lab not only to simplify the synthesis but also improve yields compared to traditional approaches.
Quinolines are used as precursors in the manufacture of antibiotics, ulcer drugs and fragrances, while thiophenes are employed in the preparation of conductive polymers used in printed circuit boards and in photographic films. In another striking example, Grzybowski and his team synthesised an anti-asthma drug using the one-pot method. The drug typically would take four consecutive synthesis and purification steps.
“Our algorithms told us this sequence could be combined into just one step, and we were naturally curious to check it out in a flask,” he said. “We performed the one-pot reaction and obtained the drug in excellent yield and at a fraction of the cost the individual steps otherwise would have accrued.”
Paper three: The third paper deals with the proposed use of Chematica to predict and monitor terrorist activity, such as syntheses that could lead to the production of chemical weapons. The platform was able to identify a number of combinations which could be used to make chemical weapons that would not be picked up using current monitoring approaches.
“Since we now have this unique ability to scrutinise all possible synthetic strategies, we can also identify the ones that a potential terrorist might use to make a nerve gas, an explosive or another toxic agent,” Grzybowski says.
The team found new ways to synthesise a lethal nerve gas using a handful of ingredients that could be picked up at any supermarket, using algorithms derived from game theory to identify the strategies that are hardest to detect by monitoring programmes, for example the use of substances such as kitchen salt, clarifiers, grain alcohol and a fertilizer.
This strategy is very different from the US government’s current approach of monitoring and regulating individual substances, according to Grzybowski.
Chematica can be used to monitor patterns of chemicals that together become suspicious, instead of monitoring individual compounds, and the developers are now working with the federal government to implement the software.
Despite the promise, critics of the system claim that the system is unlikely ever to incorporate all the laws that govern chemical reactions and misses an important element – the ‘art’ that allows scientists to take a sideways step and follow a completely new path in their thinking.
In turn, proponents argue that computers have been able to match human creativity in disciplines once thought to be beyond their reach – such as music and chess – and there is no reason why chemistry should be any different.
“Our approach will accelerate synthetic design and discovery and will optimise synthetic practice at large,” says Grzybowski.
