Phase Seperation Transfer-Learning Prediction

About PSTP

Welcome to PSTP. This application provide residue-level phase separation predictions for proteins, along with weighted phase separation propensities generated by each submodel. The Scan_PS-self model is specifically trained on self-assembly PS proteins, predicting regions that drive self-assembly. The Scan_PS-Part model is trained on partner-dependent PS proteins, predicting regions involved in interactions with PS partners. The Scan_Mix model is trained on a mixture of protein types. The figure legend also displays the weighted average predicted propensity for each method. How to Use:

1. Enter your amino acid sequence in the text area above. You can include spaces and newlines for readability.
2. Ensure that your sequence contains only the 20 standard amino acids.
3. Click 'Submit' to compute the scores and view the results.
4. You can download the results as a TSV file for further analysis.

Note: The scores are generated based on algorithms developed by our research team. For more details on the methodology, please refer to our research paper and github https://github.com/Morvan98/PSTP. If you have any questions or encounter any issues, please contact me fmf.von@sjtu.edu.cn.