OHDSI Home | Forums | Wiki | Github

Null Distribution Plot Generation

Recently, the R packages I have downloaded from ATLAS do not produce a null distribution plot. I downloaded a package that had done this previously, ran it again, and no null distribution plot was produced. Is anyone else having this issue? Thanks! @Chris_Knoll @lee_evans

Hi @mattspotnitz!

If by “null distribution plot” you mean the plot here, then yes, this is no longer generated. The new version uses the Shiny app as the primary output for all tables and figures. The Shiny app has the “Systematic error” tab that show the distribution of negative controls, but it doesn’t show the orange area. I’ve created an issue to add it, but I’m not sure when I have time to work on that.

You have all the data you need to generate the plot yourself, but it would mean writing some R code. You would need to get the estimates from the cohort_method_result.csv, join those to the negative_control.csv to only select the negative controls, and use those in the plotCalibrationEffect function of the EmpiricalCalibration package.

Hi, is this plot still produced in the shiny app? Alternatively, is there another way we can calculate the number of negative controls with a p<0.05?

t