Outlyer Removal


I'm starting to find your HCS Tools nodes more and more useful and am using them to analyze HTS data in 1536-well format.

I wondered though how to properly use the Outlyer Removal node, because isn't there a risk that it also will remove all the hits, if these were to be defined as mean + 3*SD -  in fact this is a good way of identifying them once you have filtered out positive and negative controls.

Attached an image of a plate where hits should be red, similar to the positive control in column 45. So I6 is a hit, but O38 is an outlyer that strongly affects the color scale. How would I go about removing true outlyers, but not hits - or the positive control for that matter?


Hi Evert,

good to hear that you find HSC Tools useful, thanks!

You can use the node in many ways, but its main purpose was to filter outliers out that strongly affect the axis scales or color maps in visualizations of the main data or its distributions. E.g.if you what to see if there is a plate pattern then you dont care about hits or controls being filtered out. Check the second output table of the node to see what is filtered out! Personally, I am very careful with filtering data out and try take all data along my analysis...

Your problem with the Plate Heatmap Viewer nicely displays how a single value (O38) screws up the color scheme. In the View of the Heatmap Viewer try to use View -> Outlier Handling -> smoothed. This streched the color scheme in a way that it is less sensitive to outliers. Another possibility is to use View -> Colormap -> Custom. Then you can set the color scales manually or shift the central color... Both you can do either in the global level (all plates) or on each plate that you opened individually.

I think thats easier then filtering the outliers out.




The smoothened outlier handling is indeed very useful, as it allows one to clearly see patterns on the plates. Thanks for your swift reply!