Sandbox for Deep-Learning based Computational Protein Design
Python
121
958 commits
updated Aug 14, 2026
_____________________.___.____ .____
\__ ___/\______ \ | | | |
| | | _/ | | | |
| | | | \ | |___| |___
|____| |____|_ /___|_______ \_______ \
\/ \/ \/
TRILL (TRaining and Inference using the Language of Life) is a sandbox for creative protein engineering and discovery. As a bioengineer myself, deep-learning based approaches for protein design and analysis are of great interest to me. However, many of these deep-learning models are rather unwieldy, especially for non ML-practitioners due to their sheer size. Not only does TRILL allow researchers to perform inference on their proteins of interest using a variety of models, but it also democratizes the efficient fine-tuning of large-language models. Whether using Google Colab with one GPU or a supercomputer with many, TRILL empowers scientists to leverage models with millions to billions of parameters without worrying (too much) about hardware constraints. Currently, TRILL supports using these models as of v1.10.0:
| Command | Function | Available Models |
|---|---|---|
| Embed | Generates numerical representations or "embeddings" of biological sequences for quantitative analysis and comparison. Can be Small-Molecule SMILES/RNA/DNA/Proteins dpending on the model. | ESM2, MMELLON, MolT5, ProtT5-XL, ProstT5, Ankh, CaLM, mRNA-FM/RNA-FM, SaProt, SELFIES-TED, SMI-TED |
| Visualize | Creates interactive 2D visualizations of embeddings for exploratory data analysis. | PCA, t-SNE, UMAP |
| Finetune | Finetunes protein language models for specific tasks. | ESM2, ProtGPT2, ZymCTRL, ProGen2 |
| Language Model Protein Generation | Generates proteins using pretrained language models. | ESM2, ProtGPT2, ZymCTRL, ProGen2 |
| Inverse Folding Protein Generation | Designs proteins to fold into specific 3D structures. | ESM-IF1, LigandMPNN, ProstT5 |
| Diffusion Based Protein Generation | Uses denoising diffusion models to generate proteins. | Genie2, RFDiffusion |
| Fold | Predicts 3D protein structures. | ESMFold, ProstT5, Chai-1, Boltz-2 |
| Dock | Simulates protein-ligand interactions. | DiffDock-L, Smina, Autodock Vina, Gnina, Lightdock, GeoDock |
| Classify | Predicts properties with pretrained models or train custom classifiers | CataPro, CatPred, M-Ionic, PSICHIC, PSALM, TemStaPro, EpHod, ECPICK, LightGBM, XGBoost, Isolation Forest, End-to-End Finetuning of ESM2 with a Multilayer perceptron head |
| Regress | Train custom regression models. | LightGBM, Linear |
| Simulate | Uses molecular dynamics to simulate biomolecular interactions followed by automated scoring | OpenMM, MMGBSA, ProLIF |
| Score | Utilize ESM1v or ESM2 to score protein sequences or ProteinMPNN/LigandMPNN/SCASA to score protein structures/complexes in a zero-shot manner. | COMPSS, SC |
| Workflow | Automated protein design workflows. | Foldtuning |
Check out the documentation and examples at https://trill.readthedocs.io/en/latest/index.html
1,962 followers · starred Feb 2023
Sandbox for Deep-Learning based Computational Protein Design
Python
121
958 commits
updated Aug 14, 2026
_____________________.___.____ .____
\__ ___/\______ \ | | | |
| | | _/ | | | |
| | | | \ | |___| |___
|____| |____|_ /___|_______ \_______ \
\/ \/ \/
TRILL (TRaining and Inference using the Language of Life) is a sandbox for creative protein engineering and discovery. As a bioengineer myself, deep-learning based approaches for protein design and analysis are of great interest to me. However, many of these deep-learning models are rather unwieldy, especially for non ML-practitioners due to their sheer size. Not only does TRILL allow researchers to perform inference on their proteins of interest using a variety of models, but it also democratizes the efficient fine-tuning of large-language models. Whether using Google Colab with one GPU or a supercomputer with many, TRILL empowers scientists to leverage models with millions to billions of parameters without worrying (too much) about hardware constraints. Currently, TRILL supports using these models as of v1.10.0:
| Command | Function | Available Models |
|---|---|---|
| Embed | Generates numerical representations or "embeddings" of biological sequences for quantitative analysis and comparison. Can be Small-Molecule SMILES/RNA/DNA/Proteins dpending on the model. | ESM2, MMELLON, MolT5, ProtT5-XL, ProstT5, Ankh, CaLM, mRNA-FM/RNA-FM, SaProt, SELFIES-TED, SMI-TED |
| Visualize | Creates interactive 2D visualizations of embeddings for exploratory data analysis. | PCA, t-SNE, UMAP |
| Finetune | Finetunes protein language models for specific tasks. | ESM2, ProtGPT2, ZymCTRL, ProGen2 |
| Language Model Protein Generation | Generates proteins using pretrained language models. | ESM2, ProtGPT2, ZymCTRL, ProGen2 |
| Inverse Folding Protein Generation | Designs proteins to fold into specific 3D structures. | ESM-IF1, LigandMPNN, ProstT5 |
| Diffusion Based Protein Generation | Uses denoising diffusion models to generate proteins. | Genie2, RFDiffusion |
| Fold | Predicts 3D protein structures. | ESMFold, ProstT5, Chai-1, Boltz-2 |
| Dock | Simulates protein-ligand interactions. | DiffDock-L, Smina, Autodock Vina, Gnina, Lightdock, GeoDock |
| Classify | Predicts properties with pretrained models or train custom classifiers | CataPro, CatPred, M-Ionic, PSICHIC, PSALM, TemStaPro, EpHod, ECPICK, LightGBM, XGBoost, Isolation Forest, End-to-End Finetuning of ESM2 with a Multilayer perceptron head |
| Regress | Train custom regression models. | LightGBM, Linear |
| Simulate | Uses molecular dynamics to simulate biomolecular interactions followed by automated scoring | OpenMM, MMGBSA, ProLIF |
| Score | Utilize ESM1v or ESM2 to score protein sequences or ProteinMPNN/LigandMPNN/SCASA to score protein structures/complexes in a zero-shot manner. | COMPSS, SC |
| Workflow | Automated protein design workflows. | Foldtuning |
Check out the documentation and examples at https://trill.readthedocs.io/en/latest/index.html
1,962 followers · starred Feb 2023