phenoTest and variable-wise multiple testing
1
0
Entering edit mode
Jon Manning ▴ 90
@jon-manning-5708
Last seen 7.5 years ago
Hi Bioconductors, I'm using the phenoTest package to look for associations between various continuous and categorical clinical variables and methylation values stored in an eset-like object. As suggested I apply the BH method to correct for multiple testing probe-wise. However, as I understand it this does not account for variable-wise multiple testing, and with my collaborator introducing many new variables I'm concerned this will create unnecessary spurious associations. So the question is simple: should I just do a simple column (variable)-wise bonferroni correction, after the row (probe) -wise BH correction? All pointers appreciated, Regards, Jon Manning -- Dr Jonathan Manning Bioinformatics Team Centre for Cardiovascular Science University of Edinburgh Queens Medical Research Institute 47 Little France Crescent Edinburgh EH16 4TJ United Kingdom T: +44 131 242 6700 F: +44 131 242 6782 E: jmanning at staffmail.ed.ac.uk The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.
phenoTest phenoTest • 1.0k views
ADD COMMENT
0
Entering edit mode
@james-w-macdonald-5106
Last seen 10 hours ago
United States
Hi Jon, On 1/14/2013 7:07 AM, Jon Manning wrote: > Hi Bioconductors, > > I'm using the phenoTest package to look for associations between > various continuous and categorical clinical variables and methylation > values stored in an eset-like object. > > As suggested I apply the BH method to correct for multiple testing > probe-wise. However, as I understand it this does not account for > variable-wise multiple testing, and with my collaborator introducing > many new variables I'm concerned this will create unnecessary spurious > associations. > > So the question is simple: should I just do a simple column > (variable)-wise bonferroni correction, after the row (probe) -wise BH > correction? Adding more variables to the model doesn't increase the number of tests, but doing more tests does. In other words, if you are fitting multiple models, then you are increasing the number of tests, but simply adding a new variable to an existing model doesn't increase multiplicity (as long as you only use your final model). Best, Jim > > All pointers appreciated, > > Regards, > > Jon Manning > -- James W. MacDonald, M.S. Biostatistician University of Washington Environmental and Occupational Health Sciences 4225 Roosevelt Way NE, # 100 Seattle WA 98105-6099
ADD COMMENT
0
Entering edit mode
Hi Jim, Thanks very much for the reply. phenoTest looks for univariate assocations- i.e. it tests variables one-by-one using Limma, so in this case I have many tests and I guess I will have to correct appropriately. Regards, Jon On 14/01/2013 15:09, James W. MacDonald wrote: > Hi Jon, > > On 1/14/2013 7:07 AM, Jon Manning wrote: >> Hi Bioconductors, >> >> I'm using the phenoTest package to look for associations between >> various continuous and categorical clinical variables and methylation >> values stored in an eset-like object. >> >> As suggested I apply the BH method to correct for multiple testing >> probe-wise. However, as I understand it this does not account for >> variable-wise multiple testing, and with my collaborator introducing >> many new variables I'm concerned this will create unnecessary >> spurious associations. >> >> So the question is simple: should I just do a simple column >> (variable)-wise bonferroni correction, after the row (probe) -wise BH >> correction? > > Adding more variables to the model doesn't increase the number of > tests, but doing more tests does. In other words, if you are fitting > multiple models, then you are increasing the number of tests, but > simply adding a new variable to an existing model doesn't increase > multiplicity (as long as you only use your final model). > > Best, > > Jim > > >> >> All pointers appreciated, >> >> Regards, >> >> Jon Manning >> > -- Dr Jonathan Manning Bioinformatics Team Centre for Cardiovascular Science University of Edinburgh Queens Medical Research Institute 47 Little France Crescent Edinburgh EH16 4TJ United Kingdom T: +44 131 242 6700 F: +44 131 242 6782 E: jmanning at staffmail.ed.ac.uk The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.
ADD REPLY

Login before adding your answer.

Traffic: 628 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6