bookmark OCTOPUS / obtainer of correct topologies for uncharacterized sequences Predicts transmembrane protein topology. OCTOPUS uses a combination of hidden Markov models (HMM) and artificial neural networks. In particular, OCTOPUS fully integrates modeling of reentrant/membrane-dipping regions and transmembrane hairpins in the topological grammar. It first performs a homology search using BLAST to create a sequence profile. This is used as the input to a set of neural networks that predict both the preference for each residue to be located in a transmembrane (M), interface (I), close loop (L) or globular loop (G) environment and the preference for each residue to be on the inside (i) or outside (o) of the membrane. In the third step, these predictions are used as input to a two-track HMM, which uses them to calculate the most likely topology. OCTOPUS is available online and can be downloaded as part of the TOPCONS software package.