dsmurrell / camb-paper Goto Github PK

"Additionally, two ensemble modeling approaches, namely: greedy and stacking optimization (Cortes-Ciriano et al., 2014)"

what do you think?

Introduction

(i) compound standardisation, (ii) descriptor calculation, (iii) preprocessing, model training and validation, and (iv) bioactivity/property prediction for new molecules.

Description

(i) compound standardisation, (ii) descriptor calculation, (iii) preprocessing, model training and validation, and (iv) visualisation.

Maybe we should change the introduction to fit the description?

What do you think we should do to be more consistent?

I've added this to the paper:

\subsection{Predictions for new molecules}
One of the major benefits of having all the tools available in one framework is that it makes it easy to run new molecules through exactly the same routines that the training set of molecules was subjected to before the model training process. {\it camb} provides the option to provide a test set of molecules along with a trained model, and outputs predictions on the test set.

but I don't think we have written this function actually! have we?

if not I'll add it to myself on the camb issues list.

I will need also to add my supervisor (Robert Glen). Where should he go? Usually he is always last author but since we are sharing authorship, I don't know if he will instist this time. Would you and your supervisor mind if we put him as last author?

Do we need to define them somewhere?

These were his comments:

"Can you also input minimum required numbers of atoms, and is this possible across all elements? This would be more flexible for the user
Also, ring count filters would be of practical relevance, and - if possible - a filter of undesired functional groups (such as e.g. PAINS)
How are solvents and counterions dealt with - are they simply stripped, or are there options the user can set how they are dealt with?"

Which of these, do you think we should implement (if any) before publication?

How are protonation states assigned during standardization, at ph7 or otherwise?