How were the transmembrane domains and signal peptides predicted?
For a detailed description of the methods used, refer to: Davis MJ, Zhang F, Yuan Z, Teasdale RD (2006). MemO: A consensus approach to the annotation of a protein’s membrane organization In Silico Biology 6:0037 Download article Kanapin A, Batalov S, Davis MJ, Gough J, Grimmond S, Kawaji H, Margrane M, Matsuda H, Schönbach C, Teasdale RD, RIKEN GER Group and GSL Members and Yuan Z (2003). Mouse Proteome Analysis. Genome Research, 13: 1335-1344. PubMed Abstract Signal Peptide prediction In order to address the problem of high false positive rates in signal peptide prediction, a consensus method was developed in which raw output scores for each sequence were collected from the primary prediction methods as follows: for SignalP-NN, the maximal y score (ymax) was collected, for SignalP-HMM, the probability score (sprob) was collected and for SPScan, the raw score was collected. Each of these three values was normalized by scaling the observed range to range [0,1]. A parameter called SPscor
