Analysis for only 2 replicates
4
0
Entering edit mode
Khan, Sohail ▴ 490
@khan-sohail-1137
Last seen 9.6 years ago
Dear list, I have 2 -wt and 2-mutant samples done on Affymetrix chips. Could you suggest the best way to analyze this experiment? Is averaging both replicates and picking genes based on fold change good?? Thanks for any suggestions. Sohail Khan Scientific Programmer COLD SPRING HARBOR LABORATORY Genome Research Center 500 Sunnyside Boulevard Woodbury, NY 11797 (516)422-4076
• 2.5k views
ADD COMMENT •
0
Entering edit mode
Fangxin Hong ▴ 810
@fangxin-hong-912
Last seen 9.6 years ago
try Rankprod which is based on fold change but with much more power. Fangxin > Dear list, > > I have 2 -wt and 2-mutant samples done on Affymetrix chips. > Could you suggest the best way to analyze this experiment? Is averaging > both replicates and picking genes based on fold change good?? > Thanks for any suggestions. > > > Sohail Khan > Scientific Programmer > COLD SPRING HARBOR LABORATORY > Genome Research Center > 500 Sunnyside Boulevard > Woodbury, NY 11797 > (516)422-4076 > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor > > -------------------- Fangxin Hong Ph.D. Plant Biology Laboratory The Salk Institute 10010 N. Torrey Pines Rd. La Jolla, CA 92037 E-mail: fhong at salk.edu (Phone): 858-453-4100 ext 1105
ADD COMMENT •
0
Entering edit mode
@james-w-macdonald-5106
Last seen 6 hours ago
United States
Khan, Sohail wrote: > Dear list, > > I have 2 -wt and 2-mutant samples done on Affymetrix chips. Could you > suggest the best way to analyze this experiment? Is averaging both > replicates and picking genes based on fold change good?? Thanks for > any suggestions. I would recommend using limma to analyze these data. Even though you only have two replicates, accounting for variability is better than ignoring it altogether. Best, Jim > > > Sohail Khan Scientific Programmer COLD SPRING HARBOR LABORATORY > Genome Research Center 500 Sunnyside Boulevard Woodbury, NY 11797 > (516)422-4076 > > _______________________________________________ Bioconductor mailing > list Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor -- James W. MacDonald, M.S. Biostatistician Affymetrix and cDNA Microarray Core University of Michigan Cancer Center 1500 E. Medical Center Drive 7410 CCGC Ann Arbor MI 48109 734-647-5623
ADD COMMENT •
0
Entering edit mode
@sean-davis-490
Last seen 3 months ago
United States
On 3/8/06 4:07 PM, "Khan, Sohail" <khan at="" cshl.edu=""> wrote: > Dear list, > > I have 2 -wt and 2-mutant samples done on Affymetrix chips. > Could you suggest the best way to analyze this experiment? Is averaging both > replicates and picking genes based on fold change good?? > Thanks for any suggestions. I wouldn't average if you can help it (if these are biological replictes--if they are not, then I would argue that you don't actually have any replication). You are then very much at the mercy of the outliers. I would try using limma or SAM, or one of several other methods for looking at microarray data designed with small samples in mind. That said, any results that you find are very suspect with only two replicates per condition, but if you are only interested in finding candidates for further analysis (the best you can hope to do with only four arrays), the ordering you get would, I think, be more robust using a statistical test than simple averaging. Sean
ADD COMMENT •
0
Entering edit mode
Based on my experience,SAM usually fails with only 2 replicates under each condition. Limma and RankProd both can handle 2 replicates. Fangxin > > > > On 3/8/06 4:07 PM, "Khan, Sohail" <khan at="" cshl.edu=""> wrote: > >> Dear list, >> >> I have 2 -wt and 2-mutant samples done on Affymetrix chips. >> Could you suggest the best way to analyze this experiment? Is averaging >> both >> replicates and picking genes based on fold change good?? >> Thanks for any suggestions. > > I wouldn't average if you can help it (if these are biological > replictes--if > they are not, then I would argue that you don't actually have any > replication). You are then very much at the mercy of the outliers. I > would > try using limma or SAM, or one of several other methods for looking at > microarray data designed with small samples in mind. > > That said, any results that you find are very suspect with only two > replicates per condition, but if you are only interested in finding > candidates for further analysis (the best you can hope to do with only > four > arrays), the ordering you get would, I think, be more robust using a > statistical test than simple averaging. > > Sean > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor > > -------------------- Fangxin Hong Ph.D. Plant Biology Laboratory The Salk Institute 10010 N. Torrey Pines Rd. La Jolla, CA 92037 E-mail: fhong at salk.edu (Phone): 858-453-4100 ext 1105
ADD REPLY •
0
Entering edit mode
There is an R package called 'pplr' which makes use of probe-level variance within chips. It handles small number of chips quite well. It might help. You can find it from http://umber.sbs.man.ac.uk/resources/puma/index.html. Cheers, Xuejun On Thu, 2006-03-09 at 00:34, fhong at salk.edu wrote: > Based on my experience,SAM usually fails with only 2 replicates under each > condition. > Limma and RankProd both can handle 2 replicates. > > Fangxin > > > > > > > > > On 3/8/06 4:07 PM, "Khan, Sohail" <khan at="" cshl.edu=""> wrote: > > > >> Dear list, > >> > >> I have 2 -wt and 2-mutant samples done on Affymetrix chips. > >> Could you suggest the best way to analyze this experiment? Is averaging > >> both > >> replicates and picking genes based on fold change good?? > >> Thanks for any suggestions. > > > > I wouldn't average if you can help it (if these are biological > > replictes--if > > they are not, then I would argue that you don't actually have any > > replication). You are then very much at the mercy of the outliers. I > > would > > try using limma or SAM, or one of several other methods for looking at > > microarray data designed with small samples in mind. > > > > That said, any results that you find are very suspect with only two > > replicates per condition, but if you are only interested in finding > > candidates for further analysis (the best you can hope to do with only > > four > > arrays), the ordering you get would, I think, be more robust using a > > statistical test than simple averaging. > > > > Sean > > > > _______________________________________________ > > Bioconductor mailing list > > Bioconductor at stat.math.ethz.ch > > https://stat.ethz.ch/mailman/listinfo/bioconductor > > > > > > > -------------------- > Fangxin Hong Ph.D. > Plant Biology Laboratory > The Salk Institute > 10010 N. Torrey Pines Rd. > La Jolla, CA 92037 > E-mail: fhong at salk.edu > (Phone): 858-453-4100 ext 1105 > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor
ADD REPLY •
0
Entering edit mode
Khan, Sohail ▴ 490
@khan-sohail-1137
Last seen 9.6 years ago
Thanks everyone for your advice. I have decided to use Limma, however I am not sure how to set up the design matrix for this analysis. Thanks again for your help.. -Sohail -----Original Message----- From: fhong@salk.edu [mailto:fhong@salk.edu] Sent: Wednesday, March 08, 2006 7:35 PM To: Sean Davis Cc: Khan, Sohail; Bioconductor Subject: Re: [BioC] Analysis for only 2 replicates Based on my experience,SAM usually fails with only 2 replicates under each condition. Limma and RankProd both can handle 2 replicates. Fangxin > > > > On 3/8/06 4:07 PM, "Khan, Sohail" <khan at="" cshl.edu=""> wrote: > >> Dear list, >> >> I have 2 -wt and 2-mutant samples done on Affymetrix chips. >> Could you suggest the best way to analyze this experiment? Is averaging >> both >> replicates and picking genes based on fold change good?? >> Thanks for any suggestions. > > I wouldn't average if you can help it (if these are biological > replictes--if > they are not, then I would argue that you don't actually have any > replication). You are then very much at the mercy of the outliers. I > would > try using limma or SAM, or one of several other methods for looking at > microarray data designed with small samples in mind. > > That said, any results that you find are very suspect with only two > replicates per condition, but if you are only interested in finding > candidates for further analysis (the best you can hope to do with only > four > arrays), the ordering you get would, I think, be more robust using a > statistical test than simple averaging. > > Sean > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor > > -------------------- Fangxin Hong Ph.D. Plant Biology Laboratory The Salk Institute 10010 N. Torrey Pines Rd. La Jolla, CA 92037 E-mail: fhong at salk.edu (Phone): 858-453-4100 ext 1105
ADD COMMENT •
0
Entering edit mode
On 3/9/06 8:14 AM, "Khan, Sohail" <khan at="" cshl.edu=""> wrote: > Thanks everyone for your advice. I have decided to use Limma, however I am > not sure how to set up the design matrix for this analysis. Thanks again for > your help.. Sohail, You will probably need to read the appropriate sections of the limma User guide. There is a section called "Two Groups: Affymetrix", which would seem to fit the bill here if I remember your design correctly. If you have problems, just show the code you are using and where things went wrong. Sean
ADD REPLY •
0
Entering edit mode
Dear list, I obtained a dendrogram for a set of d.e genes using hclust() with the following code: mygenes.tree<-exprs(mygenes.eset) r = cor(mygenes.tree, method=?pearson?) d = 1 - r hc = hclust(as.dist(r), method=?average?) plot(hc, labels = sampleNames) and tried to compare it with a heatmap with the same genes: heatmap(mygenes.tree, col=rbg, distfun = function(c){1 - Dist(c,method = "pearson")},hclustfun= function (d) {hclust(d, method=?average?) }, labCol= sampleNames,Colv=cov(mygenes.tree),Rowv=cov(mygenes.tree))) but the grouping of the samples on both dendrograms is different. Is it because I am using some parameters that differ between both methods? I have read the ?heatmap info and am not very sure that I understand the difference between the arguments Colv and hclustfun and whether they can be used together. Maybe I am introducing some confronting arguments without noticing. Could anyone check my code and help me on this? I am using R 2.2.0 Thanks in advance, David
ADD REPLY •
0
Entering edit mode
Le Thu, 9 Mar 2006 14:54:52 +0100 (CET) kfbargad at ehu.es a ?crit: > Dear list, > > I obtained a dendrogram for a set of d.e genes using hclust() with > the following code: > > mygenes.tree<-exprs(mygenes.eset) > r = cor(mygenes.tree, method=?pearson?) > d = 1 - r > hc = hclust(as.dist(r), method=?average?) > plot(hc, labels = sampleNames) > > and tried to compare it with a heatmap with the same genes: > > heatmap(mygenes.tree, col=rbg, distfun = function(c){1 - Dist(c,method > = "pearson")},hclustfun= function (d) {hclust(d, method=?average?) }, > labCol= sampleNames,Colv=cov(mygenes.tree),Rowv=cov(mygenes.tree))) > Hi, function Dist implement distance "centered pearson" and "uncentered pearson" with methods "pearson" and "correlation" (distance based on correlation). You should have same results with functions: 1-cor(t(x),method="pearson") Dist(x,method="correlation") Antoine. -- Antoine Lucas Centre de g?n?tique Mol?culaire, CNRS 91198 Gif sur Yvette Cedex Tel: (33)1 69 82 38 89 Fax: (33)1 69 82 38 77
ADD REPLY •

Login before adding your answer.

Similar Posts
Loading Similar Posts
Traffic: 559 users visited in the last hour
Content Search
Users
Tags
Badges
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