Fig. 4

Relative percentage heatmaps from predictive and experimental datasets corroborate residue distribution differences between TMHs from single-pass and multi-pass proteins. The residue position aligned to the centre of the TMH is on the horizontal axis, and the residue type is on the vertical axis. Amino acid types are listed in order of decreasing hydrophobicity according to the Kyte and Doolittle scale [52]. The flank lengths in the TMH segments were restricted to up to ±10 residues. The scales for each heatmap are shown beneath the respective subfigure. The darkest blue represents 0% distribution, whilst the darkest red represents the maximum relative percentage distribution that is denoted by the keys in each subfigure, with white being 50% between “cold” and “hot”. The central TMH subplots extend from the central TMH residue, whereas the inner and outer flank subplots use the database-defined TMH boundary and extend from that position. (a) TMHs from the single-pass UniHuman dataset. (b) Single-pass protein TMHs from the ExpAll dataset. (c) TMHs from the proteins of the multi-pass UniHuman dataset. (d) TMHs from ExpAll multi-pass proteins. The general consistency in relative distributions of every residue type between single-pass and multi-pass of either dataset including flank/TMH boundary selection allows us to infer biological conclusions from these distributions that are independent of methodological biases used to gather the sequences. The only residue that displays drastically differently between the datasets is cysteine in multi-pass TMHs only. The most striking differences in distributions between residues from TMHs of single-pass and multi-pass proteins include a more defined Y and W clustering at the flanks, a suppression of E and D on the inside flank, a suppression of P on the inside flank and a topological bias for C favouring the inside flank