Permeability and Efflux Prediction (GNN-MTL)

Predicts cell membrane permeability and efflux transport using a multitask graph neural network. The model simultaneously predicts four endpoints: Caco-2 efflux ratio (ER), Caco-2 apparent permeability (P_app), MDCK ER, and NIH-MDCK ER. Built with Chemprop v2.1 using a message-passing neural network (MPNN) trained on a harmonized single-laboratory dataset of over 10, 000 compounds from Caco-2 and MDCK cell-line assays.

This model was incorporated on 2026-02-23.Last packaged on 2026-03-03.

Information

Identifiers

• Ersilia Identifier: eos9cvt
• Slug: permeability-efflux-mtl

Domain

• Task: Annotation
• Subtask: Property calculation or prediction
• Biomedical Area: ADMET
• Target Organism: Homo sapiens
• Tags: ADME, Permeability

Input

• Input: Compound
• Input Dimension: 1

Output

• Output Dimension: 4
• Output Consistency: Fixed
• Interpretation: The output of this template model should be interpreted like this.

Below are the Output Columns of the model:

Name	Type	Direction	Description
caco2_er	float	high	Predicted log10 Caco-2 efflux ratio. Values >log10(2)=0.30 suggest active efflux
caco2_papp	float	high	Predicted log10 Caco-2 apparent permeability (x10^-6 cm/s). Higher means more permeable
mdck_er	float	high	Predicted log10 MDCK efflux ratio. Values >log10(2)=0.30 suggest active efflux
nih_mdck_er	float	high	Predicted log10 NIH-MDCK efflux ratio. Values >log10(2)=0.30 suggest active efflux

Source and Deployment

• Source: Local
• Source Type: External
• DockerHub: https://hub.docker.com/r/ersiliaos/eos9cvt
• Docker Architecture: AMD64, ARM64
• S3 Storage: https://ersilia-models-zipped.s3.eu-central-1.amazonaws.com/eos9cvt.zip

Resource Consumption

• Model Size (Mb): 3
• Environment Size (Mb): 7878
• Image Size (Mb): 7757.89

Computational Performance (seconds):

• 10 inputs: 40.62
• 100 inputs: 28.23
• 10000 inputs: 290.58

References

• Source Code: https://github.com/chemprop/chemprop
• Publication: https://doi.org/10.1021/acsomega.5c04861
• Publication Type: Peer reviewed
• Publication Year: 2025
• Ersilia Contributor: sergivalverde

License

This package is licensed under a GPL-3.0 license. The model contained within this package is licensed under a Apache-2.0 license.

Notice: Ersilia grants access to models as is, directly from the original authors, please refer to the original code repository and/or publication if you use the model in your research.

Use

To use this model locally, you need to have the Ersilia CLI installed. The model can be fetched using the following command:

# fetch model from the Ersilia Model Hub
ersilia fetch eos9cvt

Then, you can serve, run and close the model as follows:

# serve the model
ersilia serve eos9cvt
# generate an example file
ersilia example -n 3 -f my_input.csv
# run the model
ersilia run -i my_input.csv -o my_output.csv
# close the model
ersilia close

About Ersilia

The Ersilia Open Source Initiative is a tech non-profit organization fueling sustainable research in the Global South. Please cite the Ersilia Model Hub if you've found this model to be useful. Always let us know if you experience any issues while trying to run it. If you want to contribute to our mission, consider donating to Ersilia!