visualize_multiplot.py generates identical plots for different queries

[magnuspalmblad]

In my browser, all queries look identical in the plot. They all look like the first query. I have tried this with a new output name, 2 and 3 queries, and both visualize_multiplot.py scripts (there is still one in 'other_visualization_scripts', but this crashes).

The tables are different and correct. If I remove the first query table, the script correctly visualizes the distribution in the second.

Can you replicate this behavior?

[ReinV]

No I can't replicate this behavior. I used this script to make a plot of all neoplasm queries and I can switch between all the different plots.

Could you open console in browser (ctrl+shift+i for me in chrome) and check what is printed when switching between the queries? It should print different dataframes (arrays) for the queries.

[magnuspalmblad]

Yes, there seems to be an "Uncaught ReferenceError: key is not defined" error reported there:

eval @ VM319:27
execute @ bokeh-2.0.2.min.js:351
(anonymous) @ bokeh-2.0.2.min.js:237
emit @ bokeh-2.0.2.min.js:176
emit @ bokeh-2.0.2.min.js:176
_setv @ bokeh-2.0.2.min.js:175
setv @ bokeh-2.0.2.min.js:175
set @ bokeh-2.0.2.min.js:175
change_input @ bokeh-widgets-2.0.2.min.js:90
(anonymous) @ bokeh-widgets-2.0.2.min.js:90
VM320:24 {0.0: Array(3574), 0.0_tfidf: Array(3574), 0.25: Float64Array(3574), 0.25_tfidf: Float64Array(3574), 0.5: Float64Array(3574), …}
VM320:27 Uncaught ReferenceError: key is not defined
    at d.eval (eval at get func (bokeh-2.0.2.min.js:351), <anonymous>:27:18)
    at i.execute (bokeh-2.0.2.min.js:351)
    at d.<anonymous> (bokeh-2.0.2.min.js:237)
    at c.emit (bokeh-2.0.2.min.js:176)
    at c.emit (bokeh-2.0.2.min.js:176)
    at d._setv (bokeh-2.0.2.min.js:175)
    at d.setv (bokeh-2.0.2.min.js:175)
    at d.set [as value] (bokeh-2.0.2.min.js:175)
    at a.change_input (bokeh-widgets-2.0.2.min.js:90)
    at HTMLSelectElement.<anonymous> (bokeh-widgets-2.0.2.min.js:90)
eval @ VM320:27
execute @ bokeh-2.0.2.min.js:351
(anonymous) @ bokeh-2.0.2.min.js:237
emit @ bokeh-2.0.2.min.js:176
emit @ bokeh-2.0.2.min.js:176
_setv @ bokeh-2.0.2.min.js:175
setv @ bokeh-2.0.2.min.js:175
set @ bokeh-2.0.2.min.js:175
change_input @ bokeh-widgets-2.0.2.min.js:90
(anonymous) @ bokeh-widgets-2.0.2.min.js:90
VM321:24 {0.0: Array(7018), 0.0_tfidf: Array(7018), 0.25: Float64Array(7018), 0.25_tfidf: Float64Array(7018), 0.5: Float64Array(7018), …}
VM321:27 Uncaught ReferenceError: key is not defined
    at d.eval (eval at get func (bokeh-2.0.2.min.js:351), <anonymous>:27:18)
    at i.execute (bokeh-2.0.2.min.js:351)
    at d.<anonymous> (bokeh-2.0.2.min.js:237)
    at c.emit (bokeh-2.0.2.min.js:176)
    at c.emit (bokeh-2.0.2.min.js:176)
    at d._setv (bokeh-2.0.2.min.js:175)
    at d.setv (bokeh-2.0.2.min.js:175)
    at d.set [as value] (bokeh-2.0.2.min.js:175)
    at a.change_input (bokeh-widgets-2.0.2.min.js:90)
    at HTMLSelectElement.<anonymous> (bokeh-widgets-2.0.2.min.js:90)
eval @ VM321:27
execute @ bokeh-2.0.2.min.js:351
(anonymous) @ bokeh-2.0.2.min.js:237
emit @ bokeh-2.0.2.min.js:176
emit @ bokeh-2.0.2.min.js:176
_setv @ bokeh-2.0.2.min.js:175
setv @ bokeh-2.0.2.min.js:175
set @ bokeh-2.0.2.min.js:175
change_input @ bokeh-widgets-2.0.2.min.js:90
(anonymous) @ bokeh-widgets-2.0.2.min.js:90

[magnuspalmblad]

(The above is the error twice, I captured the errors when clicking on one query and then on another.)

[ReinV]

Just to be sure, VM321:27 you can click on here, you should see some javascript code appear, is this the line "for (key in new_data)" ?

[ReinV]

Does the same problem occur in a different browser?

[magnuspalmblad]

Yes, this is line 27:

26          // replace current data
**27          for (key in new_data) {**
28                source_data[key] = [];
29                for (i=0;i<new_data[key].length;i++) {
30                    source_data[key].push(new_data[key][i]);

[magnuspalmblad]

Edge exhibits the same behavior, but gives a different error message:

[object Object]: {0.0: Array, 0.0_tfidf: Array, 0.25: Array, 0.25_tfidf: Array, 0.5: Array...}

0: Variable undefined in strict mode
bokeh-2.0.2.min.js (175,643)

[magnuspalmblad]

IE does not show anything (white page).

[ReinV]

Could you go to the visualize_multiplot.py script, and replace line 356 - 362:

            // replace current data
            for (key in new_data) {
                source_data[key] = [];
                for (i=0;i<new_data[key].length;i++) {
                    source_data[key].push(new_data[key][i]);
                }
            }

with:

           // replace current data 
            for (column in new_data) {  
                source_data[column] = []; 
                for (i=0;i<new_data[column].length;i++) {
                    source_data[column].push(new_data[column][i]);
                }
            }

And check again please? (sorry for these edits, I'm messing up github code posting )

[magnuspalmblad]

It still does not work. Here is the new error message, complaining 'column' is now undefined:

VM56:24 {0.0: Array(3574), 0.0_tfidf: Array(3574), 0.25: Float64Array(3574), 0.25_tfidf: Float64Array(3574), 0.5: Float64Array(3574), …}
VM56:27 Uncaught ReferenceError: column is not defined
    at d.eval (eval at get func (bokeh-2.0.2.min.js:351), <anonymous>:27:18)
    at i.execute (bokeh-2.0.2.min.js:351)
    at d.<anonymous> (bokeh-2.0.2.min.js:237)
    at c.emit (bokeh-2.0.2.min.js:176)
    at c.emit (bokeh-2.0.2.min.js:176)
    at d._setv (bokeh-2.0.2.min.js:175)
    at d.setv (bokeh-2.0.2.min.js:175)
    at d.set [as value] (bokeh-2.0.2.min.js:175)
    at a.change_input (bokeh-widgets-2.0.2.min.js:90)
    at HTMLSelectElement.<anonymous> (bokeh-widgets-2.0.2.min.js:90)
eval @ VM56:27
execute @ bokeh-2.0.2.min.js:351
(anonymous) @ bokeh-2.0.2.min.js:237
emit @ bokeh-2.0.2.min.js:176
emit @ bokeh-2.0.2.min.js:176
_setv @ bokeh-2.0.2.min.js:175
setv @ bokeh-2.0.2.min.js:175
set @ bokeh-2.0.2.min.js:175
change_input @ bokeh-widgets-2.0.2.min.js:90
(anonymous) @ bokeh-widgets-2.0.2.min.js:90

[ReinV]

What about:

           // replace current data 
            for (var column in new_data) {  

[magnuspalmblad]

Yes, this works!!!

[ReinV]

For some reason the previous notation (not declaring var ...) is allowed on my pc but not on your pc (maybe it has something to do with "strict mode"?). Anyways, I will change it on our github as well.