Bug: SAM legend and plot colors can be reversed, depending on how filters are added

[danicahelb]

Looking at the WWARN data on SAM...

I first filtered Genotype or haplotype assessed = pfcrt k76 to look at the wildtype. I got this plot, which looks good:

I then went back to change my filters for Genotype or haplotype assessed and added pfcrt 76t so that i could compare the wildtype to the mutant. I got this plot, which had the colors of the legend and the plot switched:

I then cleared all filters and re-added filters for Genotype or haplotype assessed ...this time filtering both pfcrt k76 and pfcrt 76t at the SAME time. I got this plot, which once again has the colors correct:

Please make sure the legend colors match to the appropriate line in the graph, regardless of how the filters are ordered, thanks!

[dmfalke]

The issue is that the legend is ordered by subset count, while the variable’s vocabulary is ordered by filter configuration.

A solution is to augment the overlay variable such that its vocab is in the same order as the legend. I’m not yet sure where to do this. More investigation is needed.

[bobular]

I think this may be fixed by #217
I can't reproduce with SCORE S. mansoni but will double check with WWARN...

[bobular]

Confirmed. Lineplot trace labelling is correct in sam-floater-overlay-order no matter which order the genotypes are filtered in.

[dmfalke]

I may have found a hack solution that involves modifying the filters array passed to the active plot. The array is modified such that the order of values is the same as the legend items produced by the useMarkers hook. I wasn’t able to test, as my internet went out last night. I’ll test in a little bit.

[chowington]

This issue wasn't automatically closed by merging the above PR, but maybe it should have been?

[bobular]

This issue wasn't automatically closed by merging the above PR, but maybe it should have been?

Yes, thanks, will do!