PIPENN-EMB — ensemble net and protein embeddings generalise protein interface prediction beyond homology

Please do not link this web address!
For stable links, please refer to www.ibi.vu.nl/programs/ or www.ibi.vu.nl/programs/pipennemb/.

The recent major upgrade on our compute server, has lead to some configuration issues that turn out to be very hard to resolve. We are still working on this, and hope to have the server back online in the course of this month (June). We apologise again for the inconvenience, and thank you for your continued patience.

The PIPENN-EMB server offers a simple interface to our ensemble-neural-net based methods for predicting protein-protein (PPI) interaction positions from a single input sequence (Thomas et al., 2024). This web server currently offers the DNET model implementation of our PIPENN-EMB interface predictor, without the ensemble part. We are working on making the ensemble model also available here. You can input your sequence of interest, and obtain a table of predicted interface positions. Please note that runtimes are typically short, up to 15 minutes in the worst case. Depending on server load, you may be experiencing some queue waiting times as well. Code is available from the PIPENN github. The training and test sets, as well as the trained models, are available from the download page. You may also be interested in our previous method PIPENN, which although slower, also offers specific predictors for protein-nucleic acid, protein-small molecule, and conformational epitope interface position prediction.	Paste in your input sequence:   or  upload your alignment:
Please cite: David P. G. Thomas, Carlos M. Garcia Fernandez, Reza Haydarlou, and K. Anton Feenstra PIPENN-EMB: ensemble net and protein embeddings generalise protein interface prediction beyond homology bioRxiv prepint (Oct 2024).
The PIPENN web server developed and Copyright (c) by K. Anton Feenstra and Reza Haydarlou.