Skip to content

README.md

Latest commit

 

History

History
51 lines (34 loc) · 2.47 KB

README.md

File metadata and controls

51 lines (34 loc) · 2.47 KB

DeepLasso

About the project

In collaboration with the Link lab at Princeton, we developed a deep learning model to predict antimicrobial activity from the sequence of lasso peptide ubonodin. Please cite: "A High-Throughput Screen Reveals the structure-Activity Relationship of the Antimicrobial Lasso Peptide Ubonodin".

Getting started

Prerequistites

Install [Pytorch =1.8.1], [scikit-learn=1.2.], [numpy], [pandas]

Dataset

The dataset for training is under the folder /input_31th. The dataset cotains:

  • mutants: The mutation annotation of the original data
  • proteins: The lasso peptide sequences used in the embedding
  • sequence_dict: The sequence dictionary used in the lasso peptide sequence embedding
  • regression: The experimentally measured enrichment values for ubonodin
  • ssts: The secondary topology labels used in topology embedding
  • topology_dict: The secondary dictionary use in the lasso peptide topology embedding

Descriptions of folders and files in DeepLasso

  • inference contains the inference example with raw data files and the data-preprocessing scripts.
  • layers contains the multi-head attention layers used in the model
  • output contains the training log files and the output of test set files in each training epoch
  • params_trained contains the pre-trained parameters that can be used by the inference.py
  • secondary_structures_version contains the model architectures embedding the secondary topology of ubonodin (ring, loop, tail)

Step-by-step operation:

Training

  • First, run cd deeplasso, to get inside the folder, and run python preprocess.py to convert the data into word embedding.

  • Second, run python train.py. This will display training and testing results on the screen. You can also find your training and testing log file under the /output folder.

  • If you want to set up hyperparameters for your model, go to the train.main() to set the hyperparams

Running the inference

Run python inference.py > inference.output

Notably, the preprocessing "preprocess.py" and trainer "Trainer.train" code were modified from DLKcat (https://github.com/SysBioChalmers/DLKcat). Please cite: Li, F., Yuan, L., Lu, H. et al. Deep learning-based kcat prediction enables improved enzyme-constrained model reconstruction. Nat Catal 5, 662–672 (2022). https://doi.org/10.1038/s41929-022-00798-z

Developer of the DeepLasso code

Xinchun (Shone) Ran

Graduate Research Assistant

Yang Lab

Department of Chemistry

Vanderbilt University