Map stress in viewer is different from map stress given by function

[bamueh]

I have a map here where if I view it using >RacViewer(omiBeta) the stress is displayed as 2452.5, but when I output the stress using mapStress(omiBeta), it's 2175.375. Is this a bug or some feature?

[bamueh]

I can send the save file, but it's quite big, so maybe email's better than attaching it here.

[bamueh]

I had a bit more of a look around:

## Look into merging optimising weirdness
m <- read.acmap('hacked_monte_map_w_2xBetaVax_nothing_else.ace')
omi_map <- read.acmap('omi_map.ace')

# merge and reoptimise. The m map contains sera with coordinates set to NA
omiBetaReoptM <- mergeMaps(m, omi_map, method = 'reoptimized-merge', 
                           number_of_dimensions = 2, number_of_optimizations = 1500, minimum_column_basis = "none")
omiBetaReoptM <- realignMap(omiBetaReoptM, m)
ptDrawingOrder(omiBetaReoptM) <- rev(ptDrawingOrder(omiBetaReoptM))
# --> mapStress(omiBetaReoptM) = 2175.375, Stress in Viewer = 2452.5, Stress after relaxing in Viewer = 2090.72

# Merge and optimise separately
omiBetaMergedNds <- mergeMaps(m, omi_map)
omiBetaMergedNds  <- optimizeMap(omiBetaMergedNds, number_of_dimensions = 2, number_of_optimizations = 1500, 
                                 minimum_column_basis = "none")
# --> this doesn't optimise at all, and gives this error 'Error: Map contains disconnected points (points that are not connected through any path of detectable titers so cannot be coordinated relative to each other). To optimize anyway, rerun with 'options = list(ignore_disconnected = TRUE)'.'

omiBetaMergedNds <- mergeMaps(m, omi_map)
omiBetaMergedNds  <- optimizeMap(omiBetaMergedNds, number_of_dimensions = 2, number_of_optimizations = 1500, 
                                 minimum_column_basis = "none", options = list(ignore_disconnected = TRUE))
# --> mapStress(omiBetaMergedNds) = 2084.419, Stress in Viewer = 2089.7, Stress after relaxing in Viewer = 2089.7

# Merge, remove sera/ag with NA coordinates, optimise 
omiBetaMerged <- mergeMaps(m, omi_map)
omiBetaMerged <- removeSera(omiBetaMerged, c('B1617-135', 'B1617-136', 'CC1053-06', '204624651', '204624723', '204624809',
                                             '204624833', '204624870', '204922561', '204924137', '204924168', '204977154',
                                             '204626924', '204795878', '204796660', '204796723', '204922625', '204923993',
                                             '204924530', '204924559', '204924654', '204924745', '204976968', 'CSpecVir37722'))
omiBetaMerged  <- optimizeMap(omiBetaMerged, number_of_dimensions = 2, number_of_optimizations = 1500, minimum_column_basis = "none")
# --> mapStress(omiBetaMerged) = 2084.217, Stress in Viewer = 2084.217, Stress after relaxing in Viewer = 2084.217

# Remove sera/ag with NA coords, reoptimized-merge
mm <- removeSera(m, c('B1617-135', 'B1617-136', 'CC1053-06', '204624651', '204624723', '204624809',
                      '204624833', '204624870', '204922561', '204924137', '204924168', '204977154',
                      '204626924', '204795878', '204796660', '204796723', '204922625', '204923993',
                      '204924530', '204924559', '204924654', '204924745', '204976968'))
omi_map_m <- removeSera(omi_map, c('CSpecVir37722'))
omiBetaReoptM <- mergeMaps(mm, omi_map_m, method = 'reoptimized-merge', 
                           number_of_dimensions = 2, number_of_optimizations = 1500, minimum_column_basis = "none")
# --> mapStress(omiBetaReoptM) = 2164.639, Stress in Viewer = 2440.47, Stress after relaxing in Viewer = 2084.26

This seems to suggest that the maps coming out of mergeMaps when run with method = 'reoptimized-merge' aren't fully relaxed.
But I still don't understand why the stress that's displayed in the viewer would be different than the stress displayed by mapStress().

[drserajames]

I agree this is really strange behaviour.

I think the source might be the treatment of poorly connected points, ie reoptimized-merge does not remove these points which leads to higher stress. I don't know why stress is calculated differently in the viewer for the omiBetaMergedNds map.

If you send me the files, I can try to make a minimal reproducible example for Sam to work on when he's available.