PID colour scheme question

Hi Mungo

If the consensus residue has a 'tie', would you expect all of the
tied residues to get coloured?

Actually, yes - although neither would be coloured for 100% PID.

Currently they don't, because the consensus is '+' and this doesn't match.

argh. I'd bet that's been in since Jalview 1.

But it wouldn't be hard to change the behaviour so they do.

We probably ought to provide an option for that so people can have the
old behaviour if they prefer it.

I guess that could be confusing e.g.






column 1 80% consensus with P = dark blue

column 2 also has 80% of residues matching 'consensus' of [RS] - but is
this a PID of 80% or 40% or 0%?

See comment above - the scale is quantitative, so R and S both receive
40 (the value calculated for them in the consensus block).

I also need to fix my branch so '+' is shown in the consensus sequence
and not multiple residues for this case!

you do. TBH the consensus sequence shouldn't be referred to at all by
this colourscheme. The same goes for the Blosum62 one, since formally,
the average of costs to mutate each residue to the modal residues for
the column should be used to compute the colour.

Is there also a knock on for reference sequence shading ? I seem to
remember making the 'get consensus sequence' method return the reference
sequence when that was defined....



On 25/10/2016 13:49, Mungo Carstairs (Staff) wrote: