Four Annotated Sceenshots on how Tree Length Matters

The ProPhylER analysis of the Acp1 protein (top) has a deep tree with lots of substitutions per site. The ProPhylER analysis of the Ier5 protein (second from top) has far fewer substitutions per site (and a narrower phylogenetic scope). Scroll further down to get a sense of how the resolution of the predictions differs between these two proteins.

Below are interface screenshots of the same two proteins contrasted above. On top is the one with lots of tree length (subs/site), on the bottom is the one with little tree length. Highlighted is a single position, in which the alignment has only two residues (as you can see by the logo letters in the Alignment Summary). Both positions are in a region of their respective protein that evolves at an intermediate rate and that is neither particularly conserved nor particularly variable compared to the rest of the protein. (The thin red line in the Plots track shows where the highlighted residue is.)

Now look at the MAPP predictions below the mouse cursor. The protein with lots of tree length has a better dynamic range of predictions because when you have that many subs/site, observing only two residues over the entire evolutionary history of the alignment indicates strong constraint. By contrast, observing two residues in the alignment of the protein with few subs/site does not indicate particularly strong constraint, so the predictions are weaker and cover less of a range. This explanation is slightly simplified, but it highlights the fact that ProPhylER's predictions are dependent on how much underlying sequence data is available.

Last updated 8/25/08