OlorenChemEngine

Twitter account for Oloren ChemEngine.

Find us at

[OlorenChemEngine @OlorenChemEng]

"A Unified System for Molecular Property Predictions: Oloren ChemEngine and its Applications"
Whitepaper for Oloren ChemEngine out now!

https://t.co/niEAhGvo2c

[OlorenChemEngine @OlorenChemEng]

Currently, our CI supports Ubuntu 18.04 pip/conda install with/without GPU.

We plan on including Windows and Mac CI testing too.

Unfortunately, it looks like there is some specific Ubuntu 18.04 configuration where we get torch installation errors which we are still working on.

[OlorenChemEngine @OlorenChemEng]

New website :)

https://t.co/jUqeoR1mgA

Two highlights:

1. A blob map of the codebase, it's amazing how much code we've written and how far OCE has come
2. A Continuous Integration dashboard, so we can make sure all the code is properly test.

[OlorenChemEngine @OlorenChemEng]

@wpwalters Thanks for flagging this, yeah right now https://t.co/UlDp1V0zRX is hooked up directly to an internal database so the networking is a bit fussy. We are migrating that to a cloud solution (where https://t.co/jUqeoR1mgA is) which should be more "security-friendly".

[OlorenChemEngine @OlorenChemEng]

Models created using Oloren ChemEngine are best-in-class for ADME/T prediction tasks

https://t.co/qnQMDEe1bG

#AI #DrugDiscovery #ADMET #QSAR #Chemistry Benchmarked on TDC ADME/T Group

[OlorenChemEngine @OlorenChemEng]

@wpwalters so they can be visualized nicely on

https://t.co/h1eKa4iZn8

for visual inspection

[OlorenChemEngine @OlorenChemEng]

@wpwalters From https://t.co/T0WKMUvwoy, the methods used are described in the "Model Parameters" column, which can be copied in and used in code.

model = oce.create_BC(model_parameter_string)
https://t.co/zzwyplUABF(list_of_smiles, list_of_property_vals)

The parameter strings are JSON...

[OlorenChemEngine @OlorenChemEng]

Deleted and reposted with figure correction

[OlorenChemEngine @OlorenChemEng]

@predict_addict Tough, was excited

[OlorenChemEngine @OlorenChemEng]

AI chemistry folks: What cool codebases are you looking to implement/utilize?

Right now we are looking at TabPFN which seemingly promises to give super-fast tabular data models!

Interested in hearing what everyone else is looking at.

#AI #Chemistry #Chemoinformatics #AutoML

[OlorenChemEngine @OlorenChemEng]

P.S. we are always on the hunt for new cool things to bring into OCE

[OlorenChemEngine @OlorenChemEng]

using calibrated confidence intervals and probabilities as well as interpretability using counterfactual methods.

[OlorenChemEngine @OlorenChemEng]

"A Unified System for Molecular Property Predictions: Oloren ChemEngine and its Applications"
Whitepaper for Oloren ChemEngine out now!

https://t.co/niEAhGvo2c

[OlorenChemEngine @OlorenChemEng]

simplified model management and reproducibility. Using OCE, we create models which achieve superior performance on ADME/Tox prediction tasks by ensembling and integrating many different molecular property prediction methods. We include model-agnostic uncertainty quantification...

[OlorenChemEngine @OlorenChemEng]

It’s pretty cool how our server based solutions are implemented.

We serialize the entire set of objects and function calls in a script and maintain those objects mirrored in the server! We can get a big server so everything is parallelized and running super fast :)

[OlorenChemEngine @OlorenChemEng]

A demonstration edition of Oloren ChemEngine is now pip installable (virtually) without dependencies!!!

pip install olorenchemengine

This is a public, server-hosted edition for demonstration purposes. A privately hosted edition would be OCE Enterprise.

[OlorenChemEngine @OlorenChemEng]

A full (local) installation of OCE is still done via the bash script in the ReadMe.

https://t.co/qnQMDDVRXy

#AI #compchem #modeling #QSAR #GNN #machinelearning