JOSS review #3
Closed this issue · 2 comments
This issue is raised as a minor issue in context of the review process for JOSS:
Right now, the color choice of highlighted sites in dms-view depends on its numeric value in the dms-data. Like this, sites with a similar value will have an almost identical color. I think it would be nice to add an additional color scheme with more differing colors (possibly also via a drop-down menu) that can be used to highlight and discriminate sites with similar numeric values.
Thank you for raising this issue, @thorvere. We were wondering if you could clarify where you were hoping to discriminate sites with similar numeric values. Are you referring to the genome view or the protein view? If you are referring to the protein view, would you hope to have a way for users to define their own site-specific colors or for dms-view to automatically assign a categorical color scale?
We (@huddlej and I) chose the current continuous color scale to support the most common use cases for deep mutational scanning experimentalists where there is a single numerical site metric that is being mapped to a protein structure in an interactive way. Admittedly, this choice forces a tradeoff between identifying sites with divergent values and those with similar values. That said, this color scale provides multiple functions that aid data exploration. First, the colors of individual sites provide a redundant encoding of the quantitative value already encoded by the positional y axis. This type of redundant encoding helps guide and reassure the user during data exploration. Secondly, the color scale visually links the genome view with the protein view, allowing the user to track where the sites they've highlighted in the genome appear in the three-dimensional structure. The visual link by color is reinforced by the details-on-demand provided by tooltips in both genome and protein views. Finally, the continuous color scale in the protein view acts as a heatmap, allowing the user to identify regions with extreme values at a glance. The continuous color scale also allows us to dynamically assign colors to sites without additional user input or strong assumptions about the data.
Hi @skhilton,
I am referring to the protein view and I indeed think it would be nice to implement a user-specified color scale (for example for cases where neighboring sites are compared and numeric differences might only be slight). But I agree that for the majority of cases a continuous color scale suits best for the analysis of deep-mutational scanning data.