TSSUNet-MB

Identification of transcription start sites of bacterial sigma70 promoters by deep multi-task learning

TSSUNet-MB is a deep learning model developed to identify the TSSs of sigma70 promoters.

Requirements

The following packages are required to rerun the provided code.

• pytorch
• tqdm
• torchsummary

Usage

We provide two ipynb files which are separated into two different processes: training and evaluation. The weights of the TSSUNet-MB were not provided, to test and assess the model's performance, the user needs to re-train the model using the provided code.

• TSSUNet-MB-training.ipynb:

  • INPUT: Promoter sequences (located inside the data directory).
  • OUTPUT: A .pt file which store the TSSUNet-MB's weight.

• TSSUNet-MB-evaluates-plots.ipynb: Assess model's performance on the test set and generate relevant plots.

  • INPUT: Promoter sequences, model's weight file (.pt format).

License

MIT