Question: Limma and logistic regression
0
gravatar for Daniel Brewer
9.5 years ago by
Daniel Brewer1.9k
Daniel Brewer1.9k wrote:
Hello, I have a situation where I have a microarray set and a binary outcome that I want to examine. If I was just looking at one gene I would use glm with family=bionomial with the expression levels as the explanatory variable. I would like to look at the whole set of genes and for that I would normally use limma, but limma has the expression level as the response variable and so the binary outcome would be an explanatory variable. Is this an equivalent approach? Is it valid and what are the differences, especially in assumptions? Thanks Dan -- ************************************************************** Daniel Brewer, Ph.D. Institute of Cancer Research Molecular Carcinogenesis Email: daniel.brewer at icr.ac.uk ************************************************************** The Institute of Cancer Research: Royal Cancer Hospital, a charitable Company Limited by Guarantee, Registered in England under Company No. 534147 with its Registered Office at 123 Old Brompton Road, London SW7 3RP. This e-mail message is confidential and for use by the a...{{dropped:2}}
microarray cancer limma • 1.6k views
ADD COMMENTlink modified 9.5 years ago by Claus Mayer330 • written 9.5 years ago by Daniel Brewer1.9k
Answer: Limma and logistic regression
0
gravatar for Claus Mayer
9.5 years ago by
Claus Mayer330
European Union
Claus Mayer330 wrote:
Hi Dan! I am sure there are better references for this but you can find some discussion on the topic logistic regression vs t-test at this link: http://udel.edu/~mcdonald/statlogistic.html . In an ideal world you assume in a regression model that the explanatory variable is fixed by the experimenter as part of the experimental design and the response then observed in the experiment. So if you gave one group of patients a treatment, the other one a control and then observed gene expression, the natural analysis would have treatment as explanatory variable and gene expression as response. I assume in your case the binary variable is not fixed but some outcome of the study too (e.g is the treatment a success or failure). In that case you should ask yourself, what are you really trying to achieve with the study. If the aim is to predict (the probability of) the outcome by the gene expression profile, logistic regression is more appropriate. In that case it would also make sense to have a look at the literature about classification in microarray experiments etc... Best Wishes Claus > -----Original Message----- > From: bioconductor-bounces at stat.math.ethz.ch > [mailto:bioconductor-bounces at stat.math.ethz.ch] On Behalf Of > Daniel Brewer > Sent: 11 March 2010 11:42 > To: Bioconductor mailing list > Subject: [BioC] Limma and logistic regression > > Hello, > > I have a situation where I have a microarray set and a binary > outcome that I want to examine. If I was just looking at one > gene I would use glm with family=bionomial with the > expression levels as the explanatory variable. > > I would like to look at the whole set of genes and for that I > would normally use limma, but limma has the expression level > as the response variable and so the binary outcome would be > an explanatory variable. Is this an equivalent approach? Is > it valid and what are the differences, especially in assumptions? > > Thanks > > Dan > -- > ************************************************************** > Daniel Brewer, Ph.D. > > Institute of Cancer Research > Molecular Carcinogenesis > Email: daniel.brewer at icr.ac.uk > ************************************************************** > > The Institute of Cancer Research: Royal Cancer Hospital, a > charitable Company Limited by Guarantee, Registered in > England under Company No. 534147 with its Registered Office > at 123 Old Brompton Road, London SW7 3RP. > > This e-mail message is confidential and for use by the > a...{{dropped:2}} > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: > http://news.gmane.org/gmane.science.biology.informatics.conductor >
ADD COMMENTlink written 9.5 years ago by Claus Mayer330
Hi Claus, That's really great, thanks for that. The binary variable is a clinical feature which is the result of cancer, so downstream of gene expression but not fixed before hand. So this fits in with the gene expression predicting this clinical feature. Thanks Dan On 11/03/2010 4:05 PM, Claus Mayer wrote: > Hi Dan! > > I am sure there are better references for this but you can find some > discussion on the topic logistic regression vs t-test at this link: > http://udel.edu/~mcdonald/statlogistic.html . > > In an ideal world you assume in a regression model that the explanatory > variable is fixed by the experimenter as part of the experimental design and > the response then observed in the experiment. > > So if you gave one group of patients a treatment, the other one a control > and then observed gene expression, the natural analysis would have treatment > as explanatory variable and gene expression as response. > > I assume in your case the binary variable is not fixed but some outcome of > the study too (e.g is the treatment a success or failure). In that case you > should ask yourself, what are you really trying to achieve with the study. > If the aim is to predict (the probability of) the outcome by the gene > expression profile, logistic regression is more appropriate. In that case it > would also make sense to have a look at the literature about classification > in microarray experiments etc... > > Best Wishes > > Claus > >> -----Original Message----- >> From: bioconductor-bounces at stat.math.ethz.ch >> [mailto:bioconductor-bounces at stat.math.ethz.ch] On Behalf Of >> Daniel Brewer >> Sent: 11 March 2010 11:42 >> To: Bioconductor mailing list >> Subject: [BioC] Limma and logistic regression >> >> Hello, >> >> I have a situation where I have a microarray set and a binary >> outcome that I want to examine. If I was just looking at one >> gene I would use glm with family=bionomial with the >> expression levels as the explanatory variable. >> >> I would like to look at the whole set of genes and for that I >> would normally use limma, but limma has the expression level >> as the response variable and so the binary outcome would be >> an explanatory variable. Is this an equivalent approach? Is >> it valid and what are the differences, especially in assumptions? >> >> Thanks >> >> Dan >> -- >> ************************************************************** >> Daniel Brewer, Ph.D. >> >> Institute of Cancer Research >> Molecular Carcinogenesis >> Email: daniel.brewer at icr.ac.uk >> ************************************************************** >> >> The Institute of Cancer Research: Royal Cancer Hospital, a >> charitable Company Limited by Guarantee, Registered in >> England under Company No. 534147 with its Registered Office >> at 123 Old Brompton Road, London SW7 3RP. >> >> This e-mail message is confidential and for use by the >> a...{{dropped:2}} >> >> _______________________________________________ >> Bioconductor mailing list >> Bioconductor at stat.math.ethz.ch >> https://stat.ethz.ch/mailman/listinfo/bioconductor >> Search the archives: >> http://news.gmane.org/gmane.science.biology.informatics.conductor >> > -- ************************************************************** Daniel Brewer, Ph.D. Institute of Cancer Research Molecular Carcinogenesis Email: daniel.brewer at icr.ac.uk ************************************************************** The Institute of Cancer Research: Royal Cancer Hospital, a charitable Company Limited by Guarantee, Registered in England under Company No. 534147 with its Registered Office at 123 Old Brompton Road, London SW7 3RP. This e-mail message is confidential and for use by the a...{{dropped:2}}
ADD REPLYlink written 9.5 years ago by Daniel Brewer1.9k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 98 users visited in the last hour