From this week's C&EN, Rick Mullin interviews Ian Shott, the head of AMRI's European operations on his business improvement philosophies:
“Yes, I think of myself as a process engineer,” Shott said, reminiscing about his efforts to combine process development R&D with contract manufacturing at Rhodia ChiRex. Shott also challenged his peers to rewrite the organic chemistry playbook by exploring the use of special catalysts, continuous reactors, and “greener” processes. These days, however, his interest in process design reaches further upstream to basic laboratory chemistry.
“I think the whole road map from discovery to mature manufacture needs to be rethought,” Shott said. “I am quite interested in things like network analysis.” Applied in other fields of R&D, notably electronics and computers, network analysis is an engineering discipline used to predict interactions between components of a system or network. In drug discovery, for example, researchers are beginning to use chemoproteomic data to develop mathematical models for analyzing protein networks. It is a matter, he argued, of interjecting principles of engineering in an environment that is dominated by traditional chemistry.
“People have been talking about systems biology, which tries to model and predict every chemical reaction, whereas network analysis isn’t going to that level of detail,” Shott said. “It is focused more on the outcome, rather than simulating the interaction, and so it is intrinsically simpler and intrinsically cheaper, using data that are available.”
Shott sees the increased emphasis on data analysis as a revolution in drug research that will bring the chemical engineer into the process earlier. In drug discovery and development, “you have your classic paradigm in which first you have the discovery chemist, then you have the organic synthesis chemists, and then, very late in the day, you get chemical engineers involved,” Schott said. “But chemical engineers can actually be involved at any one of these stages. They’ve got the mathematics, they’ve got the statistics, and they’ve got the chemistry. If they are biochemical engineers they have biology as well. This is really a hot topic for me.”I confess (not a surprise) to some level of territoriality, but this set of statements really confuses me. I am not sure what Mr. Shott means, other than that smart engineers can help chemists do math (which, I admit, is not a strong suit of chemists.) Is that the difference? That chemistry projects need more mathematics to guide their decisions? I dunno.
Our chemical engineers did not get involved in a project unless it required a dedicated production plant to make the compound. Then we would sit down with them and discuss the process. This was only after that the company decided to build a new plant.
ReplyDeleteSometimes if we had a thorny distillation problem or reactor geometry thing we would get them involved, but generally that didn't happen too often.
It was more the crystal engineering that was important.
I think was he's saying that people get too bogged down in modeling whole systems and characterizing every interaction within that system as opposed to simply changing a variable, optimizing the variable to produce the best outcome, then moving to the next one.
ReplyDeleteWhether this is a problem or not is really not my area of expertise.
Statistical DOE has gotten a lot of hype recently, but the old-school approach of "simply changing a variable, optimizing the variable to produce the best outcome, then moving to the next one" still gets results.
DeleteFrom Curt F.:
ReplyDeleteI don't understand what this guy is saying either...but as a biochemical engineer myself, I guess I should support it. Woot woot! (?)
A few points:
ReplyDelete- The key goals of the med chemist, the process chemist, and the chemical engineer are different and not always compatible, but..
- Sharing information is good. When I saw process chemists talk to med chemists before transition to development, the transition was faster and less painful. Some companies try to repeat that with the development to manufacturing transition.
- If Ian Shott wants to use more statistics in discovery and development I can only second that. He might want to teach more statistics to med and process chemists instead using engineers as statisticians.
- Systems engineers get the "systems" thing, i.e. how complex object behave. Chemistry as a system is complex. Most chemical engineers are not systems engineers and think in more Newtonian ways (cause -> result).
To apply -omics approach to chemical development is to use chemical knowledge to predict good processes. Over the pas 60(?) years there have been many attempts to build expert systems predicting synthetic paths. Some of them work - sort of. Shott wants to take the predictive powers to a new level. Outsourcing this idea to Google might work... The main problem is the data. Most of the development data is private, and its quality is generally poor.
I think what he says is mostly a self-promoting blah to impress top management, i.e. non-chemists. To get a seat on few steering committees, he claims as a manager dude with previous process catalysis experience to have unique expertise that normal medicinal chemists lack. I worry that a former engineer who is fluent in this sort of foggy and pretentious language has apparently undergone some serious damage to his frontal lobes and is useful only in politics.
ReplyDeleteMilkshake is 100% correct. "Complexity science" is the new buzz, and it's nonsense. "We just need some feedback loops!" Moreover, he's at AMRI, and let me tell you, they love that "stuff" there as it keeps everyone from looking at the man behind the curtain....
ReplyDeleteHear! Hear! Milkshake!
ReplyDeleteWhen I was young and people talked with complexity, I thought I was dumb and they must be brilliant. Today, when I am older, I think that if they can't explain it clearly to me they have the problem. Shott seems to have the problem in this instance.
ReplyDeleteRegards,
Kilomentor