BOCTOPUS2 is a significantly improved topology prediction method for putative transmembrane beta-barrel proteins. BOCTOPUS2 outputs the location and number of the transmembrane beta-strands, the orientation (i.e. facing lipids or facing the barrel pore) of residues predicted in transmembrane beta-strands. BOCTOPUS2 is trained on full Uniprot sequences and can be used to delineate the putative transmembrane barrel region in the full input sequence. This is a major novelty of BOCTOPUS2. Most of the earlier methods are trained on PDB sequences that exclude non-barrel regions making those methods less suitable for large-scale predictions from full sequences.

BOCTOPUS2 is trained on a set of 42 full-length Uniprot sequences of transmembrane beta-barrels with a known 3D structure. It is a two-stage method, in the first stage 4 separate SVMs are used to predict the per-residue location (inner-loop, outer-loop, membrane lipid-facing, membrane pore-facing). A filter is then used to determine the putative beta-barrel region and in the second stage a hidden Markov model (HMM) is used to predict the overall topology. In contrast to BOCTOPUS, BOCTOPUS2 has a simplified HMM-architecture and requires no tuning of transition and emission probabilities, thus making it independent of the data set.

Input to the server is one or several (upto five) amino acid sequence(s) in FASTA format. The user can either paste sequences in the text-area provided, or, alternatively, upload a file containing your sequences.

The server outputs the topology predictions using BOCTOPUS2 method in plain text format. The predicted topology is also displayed graphically. Additionally, a zip folder with the input sequence, output topology (plain text and graphical representation), the position specific scoring matrix (PSSM) generated and the multiple sequence alignment used to generate the PSSM is also downloadable.