This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Statistics For Merging Different Approaches

I'm considering merging algorithms together using some statistics. For structural variants study, we have RD, RP and SR algorithms, but none are perfect. So I'm thinking if one SV is supported by several different algorithms, each producing a p-value. Then I can "merge" such p-values together as the probability of how true this calling can be.

The only method I can find right now is Fisher's method. Any recommendation or thoughts? thx

• 2,144 views • link
updated by Biostar Biostar • written by michealsmith michealsmith

Fisher's method would be my approach.

• 0 views • link
written by Zev.Kronenberg Zev.Kronenberg

Perhaps you could be a little less jargony when asking questions like this? I'm interested in algorithms, statistics, and have been doing bioinformatics for decades, and I don't think I understood a single word. A few more words would be informative, enlightening, educational, and would perhaps help more people help you :) while you help others understand something new.

• 0 views • link
written by seidel seidel

1 answer

You could also take a look at:

Supporting Online Material for A Composite of Multiple Signals Distinguishes Causal Variants in Regions of Positive Selection

While this test is for selection it combines multiple tests into one score.

Specifically look at the section:

Calculation of likelihood tables and CMS test

• 0 views • link
written by Zev.Kronenberg Zev.Kronenberg

Log in to answer this question.

More posts like this