Exercise4.17 In agenomicstudyresearcherswanttoidentifydifferentlyexpressedgeneswithin a setof m genes. Let Yj N (j,2) be theobservedexpressionlevelofthe j-th gene, j
Question:
Exercise4.17 ∗ In agenomicstudyresearcherswanttoidentifydifferentlyexpressedgeneswithin a setof m genes. Let Yj ∼N (μj,σ2) be theobservedexpressionlevelofthe j-th gene, j =1, ...,m.
Forsimplicityassumethat Yj’sareindependent(maybenotsorealisticinpractice)and σ2 is known.
Thus, weneedtotestsimultaneously H0 j : μj =0 (nondifferentiallyexpressedgene)vs. H1 j : μj ̸=0
(differentiallyexpressedgene).Evidently, H0 j will berejectedif |Yj| is large(GLRT).Applyvarious multiple testingprocedurestofindthecorrespondingthresholdsfor |Yj|’s:
1. What istheBonferroniFWERcontrollingprocedureatlevel α?
2. DerivetheHolmandHochbergprocedurestocontrolFWERatlevel α.
3. Which oftheaboveFWERcontrollingproceduresisthemostpowerful?
4. DerivetheBenjamini–HochbergFDRcontrollingprocedureatlevel α.
5. Forwhichoftheaboveprocedurestheindependenceassumptionisnecessary?
Step by Step Answer:
Statistical Theory A Concise Introduction Texts In Statistical Science
ISBN: 9781032007458
2nd Edition
Authors: Felix Abramovich, Ya'acov Ritov