Introduction
PTM-X (PTM cross-talk) is a bioinformatic software to predict the cross-talk/interplay between two post-translational modification sites within a protein, by abstracting multiple features and classfying with a naive Bayes model. The features include sequence distance, structural distance, disordered region location, residue co-evolution, and modification co-evolution. An online tool is also available at http://bioinfo.bjmu.edu.cn/ptm-x/
Installation
PTM-X is particularly easy to install with using the anaconda python distribution, which covers all packages you need.
It requires Python 2.7 or later with
Numpy, Scipy, h5py, Scikit, pylab
If the solfware version is seperated with data, please download the data and put it in the directory ./data/ by $ mv data_path/data/* PTM-X_path/data/. The PTM-X is mainly designed for Linux/Mac users by command line. For Windows users, you could simply install Cygwin , then you can use the command lines below. 1. Predict test samples
1.1 Write the file for test samples ./interface/test_samples.txt with the following format: protein residue1 PTM1 residue2 PTM2
P04637 T387 phosphorylation K382 acetylation
O95786 T170 phosphorylation K172 ubiquitination
1.2 go to ./scr/ directory, and you will see a command file "step1.0_prediction.sh", then run it by
$ sh step1.0_prediction.sh
It may take a few minutes, and then you will see the prediction in the output file: ./interface/prediction_results.txt 2. Update training samples
Note: skip this step unless you have new validated PTM cross-talk data and want to update the training set. 2.1 Update cross-talk samples ./interface/crosstalkSamples/PTM_crosstalk_Release.txt with format:
protein residue1 PTM1 residue2 PTM2 relationshipe(optional) P04637 T387 phosphorylation K382 acetylation
O95786 T170 phosphorylation K172 ubiquitination
2.2 go to ./scr/ directory and you will see a command file "step0.1_update4training.sh", then run it by
$ sh step0.1_update4training.sh
Then the training set will have been updated, see the new feature files ./interface/training_positive.hdf5 and ./interface/training_negative.hdf5 Getting started
Citation
Huang Y, et al. 2015. Mol Cell Proteomics. doi:10.1074/mcp.M114.037994
