SGenheden
commented
3 months ago Hello Philip,
Thanks for your message, and for brining up this very interesting topic.
In my opinion, protection strategies is not optimized at all with current retrosynthesis algorithms. We have seen similar behavior with other AI-driven retrosynthesis tools as well, so it is not only an issue with AiZynthFinder.
Currently, the protection strategy is incorporated with the general retrosynthesis approach, they are indistinguishable: protection and deprotection is encoded as templates (there are a lot of them) and the employment of them is left to the search algorithm. Unfortunately, we have gather more and more evidence that often the protection strategy makes little chemical sense.
We are starting to realize that we probably have to de-couple the protection strategy from the retrosynthesis - but how to do that exactly is a research project. So there will not be any solutions any time soon, I believe.
Lakshidaa
Hello, @SGenheden ，Nice to meet you again! I have another question, could you give me some advice? Thanks a lot.
when using aizynthfinder, is there a way to guide the routes genereated with reasonable functional groups or protecting groups added towards to the target mol? or, does aizynthfinder have plan to add feature like adding functional groups or protecting groups with guideline? or, currently , how can archive this purpose in aizynthfinder with current version?
I will use following two examples to state that above implementation is very needed in aizynthfinder:
Example 1: In the process of synthesizing dehydroepiandrosterone, if there are two hydroxyl groups (-OH) present in the molecule simultaneously, they are likely to undergo reactions simultaneously, leading to uncertainty in the product. To avoid this situation, protecting groups can be used to protect one of the hydroxyl groups. Common protecting groups include methyl (-CH3), ethyl (-C2H5), and propyl (-C3H7), among others. In this case, how can we specifiy protecting group in certain step?
Example 2: When examining certain reaction steps, it was found that simply adding protecting groups or other functional groups around the reaction center did not push the chemical reaction towards the target compound. If effective templates exist, how can chemical reactions be guided towards the target compound?
Do you have suggestion how can we solve these issues or advice? I am thinking the further investigation of template might be helpful for this situation. For example , some template may lead to add protecting group, others may lead to add functional group. This approach needs to decide the template first, then, based on target mol and decide to apply the template or not.
Thanks, Philip Yang