To complicate things even further, the populations tested in different datasets often overlap with each other. How can we avoid double-counting associations?
Bottom line analysis provides an answer to both of these questions. It integrates results over multiple datasets and accounts for sample overlap between datasets to generate a single p-value representing the significance of the association between a variant and a phenotype.
Now, you can access bottom line p-values for individual variants on Variant pages in the Cerebrovascular Disease Knowledge Portal as well as in the other portals of the Knowledge Portal Network: Type 2 Diabetes KP, Cardiovascular Disease KP, and Sleep Disorder KP. To view bottom line p-values, open the "associations at a glance" section of the Variant page (see an example):
Choose to view "Bottom line analysis" in the PheWAS plot, and then mouse over a point to see the p-value:
We thank our colleagues at the University of Michigan, who developed the METAL method used in this analysis. Please note that this method as instantiated in the CDKP is experimental; be sure to compare the results with those from individual datasets, and contact us with any questions.