83 implement stage 5 in dashboard

[azywot]

Added stage 5 to the dashboard:

• % ACTIVITION/% INHIBITION/Z-SCORE visualizations
• interactive Z-score adjustment
• data table with compounds and their key columns

NOTE: I think we should add one more stage for generating raports etc. (updated last stages visible here)

This is what the page looks like for now, the area plot we mentioned during the meeting is TBD

Here I used all the data we have from Primary Screening within one assay as an example:

For the future version of the plot interactivity will be retained only for compounds below/above the thresholds + stats for the area plot will be added (as it gets laggy and interactivity becomes unnecessary)

NOTE 2: For the plots I tried different layouts on a grid with the data table but ultimately decided it is the most readable like this. However, I'm not an expert when it comes to styling ngl, therefore any suggestions appreciated 🙏🙏

[Tremirre]

Nice job, implementation seems good but I have a couple of suggestions regarding the functionality and general observations:

1. Z-score range

For this task I'd suggest using two-sided slider for better UX, and since the plots are quite computationally heavy (meaning each individual change forces a long reaload) you could consider adding a recalculate or apply button that would read the slider change and then regenerate the graphs

2. Plots

All three plots are quite data-heavy and take a lot of space - they all do not fit on a single page - hence I'd recommend just showing one plot at a time with a SelectionBox that would select which visualization to show (similarily as in DrugScreeningDashboard v1.0.0 with projections)

Then you'll be able to continue our struggle to make the stage not overflow the page which is nicer for the user

3. Dash DataTable

For consistency I'd recommend you copy the styling from stage 2 datatable so that they look alike (alternating rows with different colors, different font and text size)
Furthermore we might consider if showing all 100k+ compounds in a table is really necessary - maybe we should restrict that only to the special ones

4. Questions

• Why sometimes Activation % and Inhibition % values are different? (I do not remember if that was the case or not previously)
• Why is there Z-filtering being done twice? (Once in stage 3 by @zuzg and once here)

[azywot]

Nice job, implementation seems good but I have a couple of suggestions regarding the functionality and general observations:

1. Z-score range

For this task I'd suggest using two-sided slider for better UX, and since the plots are quite computationally heavy (meaning each individual change forces a long reaload) you could consider adding a recalculate or apply button that would read the slider change and then regenerate the graphs

2. Plots

All three plots are quite data-heavy and take a lot of space - they all do not fit on a single page - hence I'd recommend just showing one plot at a time with a SelectionBox that would select which visualization to show (similarily as in DrugScreeningDashboard v1.0.0 with projections)

Then you'll be able to continue our struggle to make the stage not overflow the page which is nicer for the user

3. Dash DataTable

For consistency I'd recommend you copy the styling from stage 2 datatable so that they look alike (alternating rows with different colors, different font and text size) Furthermore we might consider if showing all 100k+ compounds in a table is really necessary - maybe we should restrict that only to the special ones

4. Questions

* Why sometimes Activation % and Inhibition % values are different? (I do not remember if that was the case or not previously)

* Why is there Z-filtering being done twice? (Once in stage 3 by @zuzg and once here)

Thanks for the comments/suggestions 😊
Ad. 1, 2 and 3 - I've added a RangeSlider as you suggested, and in the meantime put the plots to separate tabs. The same styling as in the case of a data table from stage 2 was also added. Below I'm pasting the preview:

Ad. 4

• I was actually surprised that they're mosty of the same value, I've already double checked with the pre-calculated spreadsheets we received but found only overlapping values in % ACTIVATION case (no data to compare for % INHIBITION). Great that you spotted this one, we'll have to ask about some additional data to compare and decide upon handling such cases as calculate only activation/inhibition for this assay, use this formula for activation (there are two)... etc. Have these in the back of my mind, for the future need to be taken care of.
• this is Z-score here whereas @zuzg took care of Z/Z' factors, according to Dashboard for data analysis

[zuzg]

Nice job! The page looks very good and separating plots into tabs definitely improved readability. Please make sure that after merging develop (there are some conflicts) all the pages work correctly ☺️

[azywot]

Nice job! The page looks very good and separating plots into tabs definitely improved readability. Please make sure that after merging develop (there are some conflicts) all the pages work correctly ☺️

Thanks, I've just merged and went through all the stages, everything works as expected 😊

[Tremirre]

LGTM, thanks for the reply!
We're good to merge