Secondary Structure / Membrane Protein Prediction Tools


Ali2D which takes a sequence alignment as input, performs PSIPRED and MEMSAT2 on each of the sequences and finally plots information about secondary structure, transmembrane regions and amino acids on the alignment (Java Applet).



Sensitive protein homology detection and classification of outer membrane proteins (OMP) by HMM-HMM-comparison.
The software can be downloaded here



Quick2D gives you an overview of secondary structure features like alpha-helices, extended beta-sheets, coiled coils, transmembrane helices and disorder regions. Predictions by PSIPRED, JNET, Prof(Rost), Prof(Ouali), Coils, MEMSAT2, HMMTOP, DISOPRED2 and VSL2.


External secondary structure prediction servers


Meta-server that performs a BLAST search, predicts functional motifs (Prosite) (beware of false positives), secondary structure (PHDsec), solvent accessibility (PHDacc), transmembrane helices (PHDhtm), disordered and composition biased regions, coiled coils, disulphide bridges, and more.



Secondary structure prediction using neural networks. Results are returned by e-mail. PSIPRED is one of the most popular and accurate methods around. This server also offers transmembrane topology prediction with MEMSAT2 or fold recognition with GenTHREADER.



Secondary structure prediction using neural networks. Also accepts multiple sequence alignments.


External disorder prediction servers


Protein disorder prediction based on neural networks. Best results for disorder prediction in CASP6 benchmark.



Protein disorder prediction based on neural networks. Predicts propensity for loops (no helix, no strand), hot loops (floppy loops with high temperature factors), and disordered regions (called "Remark465" in DISEMBL).



IUPred server presents an algorithm for predicting intrinsically unstructured/disordered proteins and domains (IUPs) from amino acid sequences by estimating their total pairwise interresidue interaction energy, based on the assumption that IUP sequences do not fold due to their inability to form sufficient stabilizing interresidue interactions.



Protein disorder prediction. Aligns query sequence in a sliding window with a database of known ordered and disordered sequence segments and uses a neural network to predict disorder from the alignment scores.



Protein disorder prediction, returns results by e-mail


External transmembrane protein and signal peptide predictors


Predict transmembrane (TM) helices and their topology. The method uses a hidden Markov model (HMM) that distinguishes between TM helix states, inside tails, outside tails, inside loops, and outside loops. The best fit of the query sequence with this HMM is evaluated.



Predisi is a tool for predicting signal peptide sequences and their cleavage positions in bacterial and eukaryotic amino acid sequences.



Predict transmembrane (TM) helices and their topology. The method employs a set of statistical tables (log likelihood ratios) compiled from well- characterized membrane protein data. The structural states of the model are: inside loop, outside loop, inside helix end, outside helix end, helix middle.



Predict transmembrane helices and their topology



Predict outer membrane beta-barrel proteins by a hidden Markov model method. (Compare results to HHomp, which can reliably detect outer membrane beta barrels.)



A combined transmembrane topology and signal peptide predictor. Most transmembrane prediction methods are prone to predict N-terminal signal peptides as TM helix. This server is excellent at distinguishing signal peptides from transmembrane helices. Also a good TM helix predictor.



SignalP 3.0 server predicts the presence and location of signal peptide cleavage sites in amino acid sequences from different organisms: Gram-positive prokaryotes, Gram-negative prokaryotes, and eukaryotes. The method incorporates a prediction of cleavage sites and a signal peptide/non-signal peptide prediction based on a combination of several artificial neural networks and hidden Markov models.


External GPI linkage predictors


GPI linkage prediction. Uses a rule-based algorithm for identifying the signal peptide for GPI linkers which is often wrongly predicted as transmembrane helix. GPI linked proteins are achored to the outside of the cell membrane by GPI (glycosylated phosphatidylinositol).



GPI linkage prediction. Uses Kohonen self-organizing maps.



GPI linkage prediction. Looks at amino acid type preferences near a supposed omega-site and at general physical properties.