rdkit-blog/posts/2023-06-17-overlapping-Ki-assays1

[utterances-bot]

RDKit blog - The comparative safety of combining data from Ki assays

the saga contines

https://greglandrum.github.io/rdkit-blog/posts/2023-06-17-overlapping-Ki-assays1.html

[KramerChristian]

Very interesting analysis. Yet, there are quite a number of duplicate measurements with 0.0 difference. Isn't that unrealistically low?

[greglandrum]

Very interesting analysis. Yet, there are quite a number of duplicate measurements with 0.0 difference. Isn't that unrealistically low?

@KramerChristian yep. As you discussed with your co-authors in the 2012 paper, these are probably cases where people transcribed (correctly at least) values from previous experiments. The same thing is likely also true with some fraction of the IC50 data, but I'm guessing that there's less pure duplication of data there.

I think this is probably ok for the purposes of this analysis (which is a lot less rigorous than your previous work), since I'm primarily interested in seeing how much deviation when combining separate assays for the same target in the way that I think people often do it, but I will take a look at the stats if all the 0.0 values are dropped both here and in the IC50 post.

Thanks for the feedback!