final project for 4995 deep learning class at Columbia
This project builds upon existing work: https://github.com/psipred/protein-vae
Data used for training and trained models are too big for github. Available from author upon request
mprotein_seq.txt ==> Raw protein sequences
conv_cvae_metal_gen.py ==> Convolutional VAE training script
metal_gen.py ==> Fully-connected VAE training script
seq_to_seq.py ==> reconstruct sequence, no metal-binding code, see script for usage, configured for Conv VAE
seq_to_seq.py ==> reconstruct sequence with metal-binding code, see script for usage, configured for Conv VAE
utils.py ==> helper functions
Conv CVAE metal_gen.ipynb ==> jupyter notebook, for running and testing on colab (this script has not extensively tested, might differ from actual training script/ contain errors)