Figure 5

Workflow of the method. Workflow of the method. Step 1 Retrieval of 168 transmembrane protein chains from the PDBTM database. Some of them (PDB IDs: 2b6o_A, 2nwl_B, 1zcd_A, 2dyr_A, 1jb0_F) are shown. For each chain, the PDBTM annotation for the location of the chain with respect to the lipid bilayer planes is shown. The exact definition of the different regions is given in Materials and Methods. Step 2 Fragments of sizes in the range from 3 to 14 amino acids are generated from the protein chains and classified according to the region they belong to. Step 3 For each size and region fragments are clustered according to similar hydrogen bond patterns, backbone torsion angle profiles and sequence similarity. Step 4 Sequence motifs are generated for each structural class. Step 5 Functional annotation is generated by means of GO, Swiss-Prot, Prosite, and SCOPPI databases. Step 6 Sequence motifs are filtered according to their statistical significance, their specificity for transmembrane proteins, and their novelty.