WebApp sparkLines and alignment figure/table

[Barry Grant]

I was reading about the sparkLines package (see: http://journal.r-project.org/archive/2015-1/templ-kowarik-meindl.pdf ) and thought that a shiny blast/hmmer result table with a column indicating the alignment query coverage (i.e. which portions of the query sequence are covered by the blast/hmmer alignment) similar to the Table 2 WinLose graphic in the above PDF would be useful and cool. Also a sparkLine of the -log(evalue) with a point indicating current ranking of a row could be worth trying.

Perhaps also on the ALIGN tab something similar to the output of plot.fasta() might be a useful graphic particularly if we are not rendering the entire alignment by default any more.

[Lars Skjærven]

Sure. This would be nice, and I also agree on a schematic representation of the alignment in the ALIGN tab. We might need something nicer than plot.fasta though.

[Lars Skjærven]

something like this? looks boring with the grey block though

[Barry Grant]

I like it. Gray is a bit dull but it is useful. Could we use a color scale to indicate alignment quality per position somehow. I realize we don't have the 'fit' yet from which structural deviations could come from though so it should be a sew measure. Or we could show sequence grps from the clustering???

[Lars Skjærven]

Alignment quality is a vector of length ncol(aln$ali) right? that could be nice. how do I calc that? coloring of the alignment itself only gets messy.

I've added clustering and labels now:

[Barry Grant]

There are a few ways to get at alignment quality the simplest is probably coloring by 'conservation score' for each column of the alignment. Another option might be to color gap and non-gap positions (to show what will actually go forward for PCA).

When you say coloring the alignment 'gets messy' do you mean coloring the sequences by cluster group? Is it just too many colors. I guess we could then draw boxes around the groups from the dendrogram down over the sequences?

[Lars Skjærven]

Updated the top row of red conservation bars to conservation score with a color scale from white to red.

We could add a row with gap info (perhaps with blue colors) on top of this, but this might be too much on this plot.

Coloring by residues gets too messy. I'm not sure coloring by cluster groups would be nice, but the boxes around the sequences is a good idea. I'll check it out, but we have to stop soon..

One thing, this schematic representation is fairly useless in some cases where the sequences are the same.

[Lars Skjærven]

By the way, such schematic representation of the alignment could possibly be made much fancier with some rCharts (or similar)? I have too many thing going on currently, but perhaps Shashank would consider playing with this?