Frequently Asked Questions

Server usage

1. What is PolarProtPred?
   PolarProtPred predicts the subcellular localization of transmembrane proteins in epithelial cells. Currently two prediction modes are available: apical/basolateral localization (if the submitted sequence(s) localize into polarized cells) and apical/basolateral/other (for all transmembrane proteins).
2. In what format can sequences be submitted to PolarProtPred?
   Sequences can be submitted with one letter amino acid format in FASTA format. Optionally, sequences can be uploaded in FASTA formatted file.
3. What is the reliability value next to the prediction results?
   Reliability is the output of the neural network representing the quality of the prediction. It correlates well with prediction accuracy (see the graphs at the Description menu) and can be used to determine how precise the prediction is.

Publication

1. How can I cite PolarProtPred?
   Please cite
   Laszlo Dobson, András Zeke, Gábor Tusnády
PolarProtPred: Predicting apical and basolateral localization of transmembrane proteins using putative short linear motifs and deep learning
Bioinformatics, submitted

Theory

1. What is the theory behind PolarProtPred?
   PolarProtPred uses machine learning to predict the localization of membrane proteins in epithelial cells. Since sorting is often mediated via linear motifs, we built an architecture that is capable to extract such information from protein sequences. You can find more information in the Description menu and in the manuscript.