Awesome

Genesis

Automated protein backbone refinement from a protein sketch as described in:

(1) Zander Harteveld, Joshua Southern, Michaël Defferrard, Andreas Loukas, Pierre Vandergheynst, Micheal M. Bronstein, Bruno E. Correia. Deep sharpening of topolgical features for de novo protein design. ICLR MLDD (2022).

(2) Zander Harteveld, Alexandra Van Hall-Beauvais, Irina Morozova, Joshua Southern, Casper Goverde, Sandrine Georgeon, Stéphane Rosset, Michëal Defferrard, Andreas Loukas, Pierre Vandergheynst, Michael M. Bronstein, and Bruno E. Correia. Exploring "dark matter" protein folds using deep learning. (2023).

Installation

Install libraries from the requirements file:

pip install -r REQUIREMENTS

Install PyRosetta for the structure modelling part.

Usage

Sample freely from a given FORM (a string specifying how the sketch or protein backbone should look) with set loop lengths via:

python -u sample.py \
--form A1E7.B1E7.A3E7.B3E7.B4E7.A4E7.B2E7.A2E7 \
--loops x.4.4.3.5.2.1.3.x \
--num_decoys 5 \
--wts ./data/finetune_checkpoint_500 \
--prefix designedSeq \
--out_dir ./A1E7.B1E7.A3E7.B3E7.B4E7.A4E7.B2E7.A2E7_x.4.4.3.5.2.1.3.x \
--optimizer ADAM \
--opt_iterations 101 \
--num_recycling 0 \
--polyVAL_seq False \
--pssm_design

Or, use an example Google Colab Notebook to run Genesis: