rahmanszsaleem
commented
2 months ago I see a paper by Llompart and coworkers has been published after this study was completed. Their work investigated 20 years of published solubility data and concluded "Our results demonstrate that some state-of-the-art models are not ready for public usage because they lack a well-defined applicability domain and overlook historical data sources." A conclusion not very different from ours. Stegemann has recently published on the trends in oral small molecules exceeding the Rule of 5. There is more work on solubility prediction of molecular complexes by Yu and coworkers. Similarly, Liu and coworkers have published a study on solid–liquid equilibrium solubility and solvent effects of N-benzyloxycarbonyl-l-serine in 12 monosolvent systems. I think we should reach out to these authors and explore if they can contribute to improving the understanding fo the solubility.
Here are the details of what has been going on in this project, and you can find some relevant literature and background studies here.
You can participate in the following ways: 1- You can take our data and run your algorithms/software and tell if your algorithm/software gives a better correlation 2- Give us your solubility dataset for analysis. 3- Tell us if our conclusion is surprising to you or if you have found any accurate tools to use.
We wonder if you have any suggestions for what we can do for better solubility prediction.
In terms of aggregation, we will analyze the data and post the results in this repository upon receiving them.
CNP15 Solubility Reach Out.pptx