Fragment Ensemble Scoring (FRESCO)
This repo contains code describing Fragment Ensemble Scoring (FRESCO), a method for hit-finding from a fragment screen. FRESCO uses unsupervised machine learning to model a distribution of pharmacophore combinations in 3D space, which is then used to conduct a virtual screen on the EnamineREAL library for discovery of moelcular hits. This method is unique in directly predicting hit compounds from a fragment screen without any assaying of binding activity required, and has demonstrated success computationally on a retrospective analysis of COVID Moonshot compounds as well as in prospectively discovering hit compounds for both SARS-CoV-2 Mpro and SARS-CoV-2 nsp3-Mac1.
The repo contains all of the code necessary for fitting a model on fragment screen data and scoring molecules, but not the EnamineREAL datadue to storage limitations. Basic README files are present and full turotials for using FRESCO are work-in-progress!
Workflow
flowchart TD
subgraph "FRESCO Workflow"
subgraph EnamineREAL
A[(VirtualFlow\nConformers)] --> B(Pharmacophore Descriptors)
end
subgraph Ensemble of Fragment-Protein Complexes
Mpro[(Mpro)] --> PharmD(Pharmacophore Distributions)
Mac-1[(Mac-1)] --> PharmD
PharmD -- scikit-learn + scipy interp1d--> KDE(KDEs of pharmacophore distances)
end
subgraph Virtual Screening
B --> E(Score Library)
KDE --> E
E --> F(Choose Top-50k)
F --> G(Filtering)
G -- "physchem (lead-like)"--> H(Clustering)
G -- PAINS --> H
G -- "at most one chiral center" --> H
G -- remove duplicate tautomers --> H
end
end
Usage
FRESCO functions can be used as a python library by importing the fresco module via pip install .* in the root directory of the repo.