Some questions about the statistical method

[Jingxian-Ren]

Dear professor: Good afternoon. After we use CASCADE to predict chemical shifts for several weeks, I have some questions about the statistical method used for processing data. Could you please help with these questions?
First, if we use the CASCADE to verify that the resolved compound structure is correct, approximately how much MAE can prove the correctness of the structure? Secondly, when we use MAE to determine the structures, sometimes the MAE worked wrongly. So can CASCADE be used with DP4 etc.? Looking forward to your answer. Thanks a lot.

[bobbypaton]

Thanks for the questions. Firstly, the expected 13C MAE of a correct structure, based on the molecules we studied in the publication is 1-2 ppm. For a large database of natural products 13C NMR spectra measured in various solvents, perhaps a little larger. How much worse an incorrect structure will be depends on how it differs from the correct one – i.e. structural vs. stereochemical difference. In our publication we saw examples where the incorrect structure was around 1ppm worse MAE for structural revisions, and around 0.5 ppm worse MAE for stereochemical revisions. This is, however, likely to be system dependent. Finally, yes we could and probably should implement the DP4 metric, although this hasn’t been done yet. Note that this only really makes sense for stereochemical comparisons or cases where you know that one of your candidate structures must be correct. Rob Robert Paton | Professor | Colorado State University http://patonlab.colostate.edu<http://patonlab.colostate.edu/> From: Jingxian-Ren ***@***.***> Date: Wednesday, February 16, 2022 at 12:54 AM To: patonlab/CASCADE ***@***.***> Cc: Subscribed ***@***.***> Subject: [patonlab/CASCADE] Some questions about the statistical method (Issue #5) ** Caution: EXTERNAL Sender ** Dear professor: Good afternoon. After we use CASCADE to predict chemical shifts for several weeks, I have some questions about the statistical method used for processing data. Could you please help with these questions? First, if we use the CASCADE to verify that the resolved compound structure is correct, approximately how much MAE can prove the correctness of the structure? Secondly, when we use MAE to determine the structures, sometimes the MAE worked wrongly. So can CASCADE be used with DP4 etc.? Looking forward to your answer. Thanks a lot. — Reply to this email directly, view it on GitHub<https://nam10.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fpatonlab%2FCASCADE%2Fissues%2F5&data=04%7C01%7Crobert.paton%40colostate.edu%7C08589f72e6fd4658b66708d9f1218fcf%7Cafb58802ff7a4bb1ab21367ff2ecfc8b%7C0%7C0%7C637805948720631611%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=OREWojvplgbUOUI2QoSU7j8QVS3jKduQUfbXMaJoFg4%3D&reserved=0>, or unsubscribe<https://nam10.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAACOQJKL2QC45XTJWHTXJFLU3NJ3LANCNFSM5OQZ5OJA&data=04%7C01%7Crobert.paton%40colostate.edu%7C08589f72e6fd4658b66708d9f1218fcf%7Cafb58802ff7a4bb1ab21367ff2ecfc8b%7C0%7C0%7C637805948720631611%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=VfB3J5b%2BwEf8kyxCfCCp5iCQwrvi9AlzfUx1cgH%2B8Y8%3D&reserved=0>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>