Abstract: “1,3,4-oxadiazole plays a crucial role in organic synthesis and medicinal chemistry due to its broad-spectrum of bioactivity as an anticancer, antimicrobial, antifungal, and antifungal pharmacological agent. However, traditional approaches of multicomponent synthesis for the synthesis of 1,3,4-oxadiazole scaffold have a few limitations such as multistep synthesis, limited substrate scope, or use of harsh reaction conditions. Among reported synthesis of 1,3,4-oxadiazole, one strategy is using four building blocks to get 1,3,4-oxadiazole in a one-pot manner utilizing multicomponent reaction approach as reported by Ramazani et al and Yudin et al. One of the key steps is an imine formation, that forms in situ from aldehyde and amine, which could be partially reversible depending upon the substrate used as well as reaction conditions. In order to overcome the limitation of the previous method, we generate a machine learning model that can interpret the chemical reactivity and reaction outcome of the desired product.”
East Central IL ACS Undergraduate Research Conference
Thank you for joining us for the 5th Annual ECI-ACS conference!
Very interesting presentation. Could you explain what parameters were used in the development of the machine learning model? Additional DFT simulations were mentioned in the poster, could you provide some additional detail on how that was used in the development of the machine learning model?
Hi Steven, I worked mostly on the synthetic part. If you are interested in the computational part, please see our supporting information of reference 1. Thank you.
Hi Yingqi, I enjoyed your presentation! You state that one of the primary challenges with generating a machine learning system from existing research is that most literature only reports successful results. Would you be able to go into more detail about how you populate your algorithms with “unsuccessful” approaches and how you can obtain the same breath of “unsuccessful” approaches to adequately balance out against the successful approaches that have been previously reported? Also, what is the outlook for future models? Ideally, yields greater than 70% are desired, so has your research shown any promising routes to get even higher yields?
Since the database containing unsuccessful results are not usually available, in our work, we work on a case study where sulfonylimine is a model substrate. We had a close to a 1:1 ratio of successful and unsuccessful reactions. The current model doesn’t involve yield information as it was for reactivity prediction. I agree that high yield is beneficial but our prime focus was on reactivity rather than yield. Future prospective will be on reaction conditions and yields.
And I work mostly on the synthetic part.
Thanks!
No problem:)