This paper in PLOS ONE shows a disturbing lack of correlation between IC50’s for the Ephrin type-B receptor 4 (EphB4) when the compounds were dispensed by conventional disposable pipet tips and acoustic dispensing. The R2 was 0.25. Also catalyst pharmacophores developed with the two datasets were visible different, which isn’t surprising when you consider the lack of correlation between the two dispensing methods. The really disturbing part is that when one takes receptor pharmacophores generated from the kinase’s ligand/enzyme complex crystal structures, they agree with the acoustic dispensing data pharmacophore, and are different than the disposable tip generated data.
I think we have all been aware of the issues with compounds liking to stick to plastics, but this paper suggests a reason why all the data we have been generating and trying to model is so lacking in predictive value. I personally wonder how many times models I have used to try to help drive a project were doomed before I began working on the project, because the data was contaminated with sticky plastic tips. This adds one more factor to check when one is building models. A very important factor. I hope we can see more studies like this and determine if a mass flight from disposable tips to acoustic dispensing is necessary.
I think we have all been aware of the issues with compounds liking to stick to plastics, but this paper suggests a reason why all the data we have been generating and trying to model is so lacking in predictive value. I personally wonder how many times models I have used to try to help drive a project were doomed before I began working on the project, because the data was contaminated with sticky plastic tips. This adds one more factor to check when one is building models. A very important factor. I hope we can see more studies like this and determine if a mass flight from disposable tips to acoustic dispensing is necessary.
RSS Feed
