EpHod is a deep-learning model to predict the optimum pH of enzymes (pHopt). The model architecture consists of a light-attention module with 49 million parameters on top of the protein language model, ESM-1v, with 690 million parameters. EpHod was trained on 1.9 million proteins with optimal environment pH (pHenv) followed by 9,855 enzymes with of catalytic optimum pH (pHopt).
We recommend using a conda environment.
Dependencies are in env.yml.
Weights of EpHod model and training datasets are available at Zenodo.
1. Clone repository and install conda environment. Installation with the required environment takes roughly four minutes.
git clone https://github.com/jafetgado/EpHod.git
cd EpHod
export PYTHONPATH="$(pwd)"
conda env create -f ./env.yml -p ./env
conda activate ./env
2. Predict pHopt with EpHod (needs gpu). Predicted pHopt values, and
softmax weight values (attention weights), as well as final EpHod layer embeddings
(2560-dim) are saved in ./example/.Pass 0 to --save_attention_weights
and --save_embeddings to avoid writing the weights and embeddings output.
Besides downloading model weights, which may take several minutues, with a batch
size of 1, prediction takes ~7 seconds/sequence on a CPU and ~0.1 seconds/sequence
on a GPU.
python ./ephod/predict.py \
--fasta_path "./example/sequences.fasta" \
--save_dir ./example \
--csv_name ephod_pred.csv \
--batch_size 1 \
--verbose 1 \
--save_attention_weights 1 \
--attention_mode "average" \
--save_embeddings 1
3. Alternatively, predict pHopt with a support vector regression model based on the amino acid composition (AAC-SVR). This may be less accurate but is a very fast estimation with CPU.
python ./ephod/predict.py \
--fasta_path "./example/sequences.fasta" \
--save_dir ./example \
--csv_name svr_pred.csv \
--aac_svr 1 \
--verbose 1
If you find EpHod useful, please cite the following:
Gado J.E., Knotts M., Shaw A.Y., et al, 2023. "Deep learning prediction of enzyme optimum pH". bioRxiv.