API and Docs

---

API

The API for the model can be accessed via this link.

Source code and steps for generation are available on GitHub.

Model Details

---

Overview

• Conformal regression model for predicting LogD for drug-like molecules, trained on ChEMBL version 33 with over 2.2M compounds using the CPSign software.
• This is a re-take of the cpLogD model previously published in A confidence predictor for logD using conformal regression and a support-vector machine. The main update is that it is based on a newer version of ChEMBL (v33, May 2023), and that it is built using a later version of CPSign - which is now open source for non-commercial use. Note: the old cpLogD was based on the computed property acd_logd in ChEMBL but which is no longer supplied, and have been replaced by the CX LogD 7.4 property which was now used.

Version

Owners

• Staffan Arvidsson McShane