DeepSig is a software package for the prediction of signal peptides ad their cleavage sites on protein sequences.
The source code and installation instructions are available on GitHub at https://github.com/BolognaBiocomp/deepsig
If you are familiar with Docker, we also provide a Docker image which allows to easily run DeepSig using Docker.
The image as well as instructions on how to use the containerized version of DeepSig are available on DockerHub at https://hub.docker.com/r/bolognabiocomp/deepsig
This dataset has been constructed to train/test the SignalP4.1 predictor. It comprises 7760 Eukaryotic, 685 Gram-positive and 1858 Gram-negative protein sequences, respectively. Each organism datasets contains posivite proteins endowed with signal peptides as well as two different negative sets: proteins with a transmembrane region annotated in the first 70 residues and nuclear/cytosolic proteins. Redundancy has been reduced internally to 25% sequence identity.
This dataset was generated to compare DeepSig with other available methods. It contains 1058 Eukaryotic, 438 Gram-positive and 211 Gram-negative protein sequences, respectively. Each organism datasets contains posivite proteins endowed with signal peptides as well as two different negative sets: proteins with a transmembrane region annotated in the first 70 residues and nuclear/cytosolic proteins. Redundancy has been reduced both internally as well as with respect to the SignalP4.0 dataset to 25% sequence identity.