Treatment of Affinity Measurements in Original PLEC Paper

[mate...@gmail.com]

Hello,

In the original published paper regarding the PLEC fingerprint, it's stated that training was done on the PDBbind general set. Looking at the data, it seems that about 1/3 of the complexes have their affinity measurements as IC50 values, while the remainder are in Ki/Kd.

My question is about the methodology used with regards to IC50 vs Ki/Kd measurements. Were IC50 values treated the same as Ki/Kd values or was some conversion applied before feeding them into the ML model?

For example, the following paper states that a conversion factor of 2 is most suitable for mixing IC50 and Ki/Kd data:

https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0061007

Thank you in advance,

Mateo Vacacela

[Maciek Wójcikowski]

Hi,

No additional processing was done on the target values in the PLEC paper. I also have not conducted any study if it helps at all, but it seems like an interesting problem. 

--
You received this message because you are subscribed to the Google Groups "Open Drug Discovery Toolkit Community" group.
To unsubscribe from this group and stop receiving emails from it, send an email to oddt+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/oddt/f17bba77-5eac-4164-8999-a5ec4b92d74e%40googlegroups.com.