MultiLoc

ChloroP is based on neural networks and the recognition of chloroplast transer peptides (cTPs). The method discriminates between chloroplast and non-chloroplast proteins.
iPSORT is based on rules, which use physiochemical properties like hydrophobicity to discriminate between N-terminal targeting sequences (cTPs, mTPs, and SPs).
MitoProtII is based on discrimant function analysis and calculates several physiochemical parameters obtained from literature. The method distinguishes between mitochondrial and non-mitochondrial proteins.
NNPSL is based on neural networks and overall amino acid composition. The method discriminates between three possible locations (cytoplasmic, extracellular, and periplasmic) for procaryotes and for locations (cytoplasmic, extracellular, mitochondrial, and nuclear) four eukaryotes.
PLOC was trained using support vector machines and is based on the overall amino acid composition, amino acid pair composition, and gapped amino acid pair composition. The method discriminates between 12 subcellular locations.
PredictNLS is a method for predicting nuclear localization signals (NLS) and it is based on NLSdb, a database of experimentally known or potential mono- and bipartite NLS motifs.
Predotar is based on neural networks and discriminates between N-terminal targeting sequences (cTPs, mTPs, and SPs).
PSORT consists of a series of if-then rules collected from literature, which use the results of several subprograms for the final prediction. The method discriminates between 17 subcellular locations.
SignalP is a method for distinguishing between signal peptides (SPs), signal anchors (SAs), and non-secretory proteins. There are two versions of SignalP. One is based on neural networks and the other on hidden Markov models.
SubLoc is based on support vector machines and was trained using the NNPSL training data set and therefore predicts the same locations.
TargetP is based on neural networks and discriminates between N-terminal targeting sequences (cTPs, mTPs, and SPs).