Definition of hydrophobic and loop clusters. The sequence of the C-terminal end of the phospholipase C δ PH domain, whose structure has been experimentally solved (pdb 1mai), is shown as example. The sequence is translated into two binary sequences, using on the one hand, a hydrophobic cluster specific-code (1 = VILFMYW, * = P, 0 = other residues) and on the other hand, a loop cluster specific-code (1 = PGDNS, 0 = other residues). The consensus assignment of secondary structures is shown between the two binary sequences (E = strand, H = helix). On the binary-encoded sequences, hydrophobic clusters are shaded in green (1 = VILFMYW), whereas PGDNS and loop clusters are colored in grey and blue, respectively (1 = PGDNS). Loop clusters are necessarily included in PGDNS clusters (the blue color superimposes with the grey one), as information brought by the hydrophobic clusters has been omitted from the PGDNS clusters to form loop clusters. Loop clusters and hydrophobic clusters are therefore non-intertwined. Hydrophobic clusters and loop clusters always begin and end with "1". The positions encoded by 1 within clusters (hydrophobic, PGDNS and loop clusters) are colored accordingly on the 2D HCA representation, shown at right. The Peitsch codes (P-codes) of the two hydrophobic clusters are indicated at bottom. Note that the α-helix is longer than the corresponding hydrophobic cluster (6 amino acids upstream). However, this sequence includes two alanine residues, which have strong propensities for helical structure (e.g. ). Alanine has not been integrated into the hydrophobic alphabet, as it does not increase the global mean correspondence between hydrophobic clusters and regular secondary structures (data not shown). This amino acid has to be considered in a context-dependant way. HCA plots were drawn using DrawHCA .