code for paper IsEM-Pro: Importance Weighted Expectation-Maximization for Protein Sequence Design
conda create -n isem-pro python=3.8 -y
conda activate isem-pro
conda install pytorch=1.10.2 cudatoolkit=11.3 -c pytorch -y
conda install numpy=1.19 pandas=1.3 -y Clone this repository by the following commands:
git clone git@github.com:JocelynSong/IsEM-Pro.git
cd IsEM-Promkdir avGFP
bash run_mrf.sh avGFPbash train_vae_mrf.sh data_path avGFP outout_pathThen train the latent generative model using MCEM:
bash train_is_vae.sh data_path avGFP outout_path first_stage_pathThe sample number (--max-iteration-sample) is 10% of the original training data size.
bash generate_vae.sh data_path avGFP outout_path generation_pathgeneration_path/protein.txt is the final output file