Meeting 8 - GeorgeIniatis/Blood_Brain_Barrier_Drug_Prediction GitHub Wiki

Meeting Minutes

Mainly discussed the PCA and TSNE visualisations produced

I mentioned that the plots seem to indicate that there is too much overlap between the 2 classes, but the supervisor repeated that strong conclusions cannot be drawn from these plots.

I suggested adding another dimension to the plots and trying UMAP

The supervisor then mentioned that he has prepared some code that I could potentially use to retrieve some more compounds for the dataset if it is not too much trouble

Action Plan

• Produce a simple classifier to discuss in the next meeting
• Try to improve visualisations and try UMAP
• Have a look at the material sent over by the supervisor