DeepPSP

DeepPSP provides the prediction of general and kinase-specific phosphorylation site using deep leraning. Developer is Lei Guo from Laboratory of Biomedical Network, Department of Electronic Science, Xiamen University of China.

Requirement

python == 3.5, 3.6 or 3.7

keras == 2.1.2

tensorflow == 1.14.0

numpy >= 1.8.0

backend == tensorflow

Predict For Your Test Data

cd to the DeepPSP fold

If you want to predict general site, taking S/T site as a example, run:

python predict.py -input ...\DATASET\test_general_ST.fasta -predict-type general -output ...\result_general_model_ST -residue ST

If you want to predict kinase-specific site, taking MAPK as a example, run:

python predict.py -input ...\DATASET\test_MAPK.fasta -predict-type kinase -kinase MAPK -output ...\result_kinase_model_MAPK -residue ST

Output file includes three columns, position, residue type and score. The value range of score is [0, 1], with values closer to 1 indicating the site is more likely to be phosphorylated.

Train For Your own Data

“#” should be first added after each phosphorylation site

if you want to train for general site, then run:

python train.py -input ...\DATASET\train_general_ST###.fasta -train-type general -residue ST   

if you want to train for kinase-specific site, taking MAPK as a example, then run:

python train.py -input ...\DATASET\train_MAPK###.fasta -train-type kinase -kinase MAPK -residue ST   

Note that pre-training model would been loaded before kinase-specific site training. If you don't need it, you can train model on general site.

Contact

Please contact me if you have any help: gl5121405@gmail.com

Citation

Guo L , Y Wang, Xu X , et al. DeepPSP: A Global–Local Information-Based Deep Neural Network for the Prediction of Protein Phosphorylation Sites[J]. Journal of Proteome Research, 2020, 20(1).