Closed dvmartinweigl closed 11 months ago
Reopening this since the PR #119 closed it automatically
Raincloud plot makes application crash if we set a column with null values only (here EC50) and then change the cloud options
This is not reproducible by me -
Reopening this since the PR https://github.com/datavisyn/visyn_core/pull/122 closed it automatically
In bioinsight we are facing errors within some vis types caused by null values in the data (we have a sparse matrix). Some components (hexbin plot) even make the whole application crash if invalid data got passed. Unfortunately it is very hard to check and filter out the null values, because this is based on the current vis config.
@thinkh , @dvmoritzschoefl and me decided to handle the data cleansing in the vis components itself
Here is a list of some null value errors and other small bioinsight related bugs we observed:
[x] Bar char fails in bioinsight if
multiples
column got selected (seems to only happen in production build, to reproduceyarn run webpack:prod && yarn start dev_server_only=true
)[x] Violin plot has misaligned or missing x axis labels after changing the categorical column. Changing the mosaic split size makes the labels render correctly.
[x] Raincloud plot actually just needs 1 numerical column. Please change the invalid settings message accordingly.
[x] Raincloud plot makes application crash if we set a column with null values only (here EC50) and then change the cloud options
[x] Corrialtion plot: change invalid settings text to select at least 2 numerical columns
[x] Scatter plot: fails on sparse matrix with columns > 2
[x] Hexbin plot: application crashes if null values are present in data
[x] Hexbin plot: hide legend if invalid settings
[x] Correlation plot: cannot handle integer values