quality data check
2
0
Entering edit mode
@marcelo-luiz-de-laia-377
Last seen 7.1 years ago
Dear All, We work with cDNA in nylon membranes. We use bioconductor to analyze our data, basically, the packages limma, affy and vsn. The normalizations are made using normalize.quantiles from affy package. However, to verify the quality of the data from this platform is a little complicated with the packages bioconductor. We don't know how to do, if it is possible. Soon, we made one boxplot for each membrane (36 membranes). They are 6 treatments with 6 biological repetitions. Each gene was printed 2 times in the membrane. The raw data had been transformed for log into base 2. We are with doubt in relation to the interpretation of the graphs. [not normalized] http://www.lbm.fcav.unesp.br/download/todos.png [normalized] http://www.lbm.fcav.unesp.br/download/todos.norm.png They are: 1. it has outliers. The presented amount is very much or is relatively normal for this platform? 2. the box is above of line zero (y axis). That is normal? After normalized the data, they are between position 2 and the 3 (y axis). This can happen? We intend to pick up learn with yours commentaries, therefore, you are free to send us any others commentary about the data or quality check on this platform. Yours suggestions is very apreciated. Thanks very much Marcelo
affy vsn limma affy vsn limma • 580 views
ADD COMMENT
0
Entering edit mode
@gustavo-henrique-esteves-937
Last seen 6.3 years ago
Brazil
Hello Marcelo, it seems to me that your data wasn't normalized for dye bias. in general, when your data is well normalized, the mean (median) expression values remains around zero... I'm not familiar with affy package, but try to look at 'normalize' function from this package... []s Gustavo 2005/5/12, Marcelo Luiz de Laia <mlaia@fcav.unesp.br>: > > Dear All, > > We work with cDNA in nylon membranes. We use bioconductor to analyze our > data, basically, the packages limma, affy and vsn. The normalizations > are made using normalize.quantiles from affy package. > > However, to verify the quality of the data from this platform is a > little complicated with the packages bioconductor. We don't know how to > do, if it is possible. > > Soon, we made one boxplot for each membrane (36 membranes). They are 6 > treatments with 6 biological repetitions. Each gene was printed 2 times > in the membrane. > > The raw data had been transformed for log into base 2. > > We are with doubt in relation to the interpretation of the graphs. > > [not normalized] > http://www.lbm.fcav.unesp.br/download/todos.png > > [normalized] > http://www.lbm.fcav.unesp.br/download/todos.norm.png > > They are: > > 1. it has outliers. The presented amount is very much or is relatively > normal for this > platform? > > 2. the box is above of line zero (y axis). That is normal? After > normalized the data, they are between position 2 and the 3 (y axis). > This can happen? > > We intend to pick up learn with yours commentaries, therefore, you are > free to send us any others commentary about the data or quality check on > this platform. Yours suggestions is very apreciated. > > Thanks very much > > Marcelo > > _______________________________________________ > Bioconductor mailing list > Bioconductor@stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor > -- _______________________________________________________________ "The truth you speak has no past and no future. It is, and that's all it needs to be." "A verdade que vocę fala năo tem passado nem futuro. Ela é, e isso é tudo que ela precisa ser." Autor desconhecido _______________________________________________________________ Gustavo Henrique Esteves e-mail: gesteves@gmail.com Home Page: http://www.vision.ime.usp.br/~gesteves/ [[alternative HTML version deleted]]
ADD COMMENT
0
Entering edit mode
Hi Marcelo, Gustavo, Gustavo H. Esteves wrote: > Hello Marcelo, > it seems to me that your data wasn't normalized for dye bias. I assume you use radioactive labeling on the cDNA membranes - so there are no dyes. Your boxplots look fine to me, almost too clean - the few outliers at the bottom and top are perfectly normal, and the quantile normalization has indeed achieved its goal of making all distributions look the same. As Sean suggests, try looking at the scatterplots. Also, if "x" is your exprSet or AffyBatch with the data, try the following plot: cols = colorRampPalette(c("blue", "yellow"))(256) cc = cor(exprs(x)) image(cc, col=cols) With this you can find outlier arrays - they are blue. ----------------- To assess the performance of the normalization, you could calculate the F-statistics for each gene (e.g. using "rowFtests" from genefilter package) and compare them before and after normalization, or between different ways of normalizing / doing quality control. Best wishes Wolfgang > > 2005/5/12, Marcelo Luiz de Laia <mlaia@fcav.unesp.br>: > >>Dear All, >> >>We work with cDNA in nylon membranes. We use bioconductor to analyze our >>data, basically, the packages limma, affy and vsn. The normalizations >>are made using normalize.quantiles from affy package. >> >>However, to verify the quality of the data from this platform is a >>little complicated with the packages bioconductor. We don't know how to >>do, if it is possible. >> >>Soon, we made one boxplot for each membrane (36 membranes). They are 6 >>treatments with 6 biological repetitions. Each gene was printed 2 times >>in the membrane. >> >>The raw data had been transformed for log into base 2. >> >>We are with doubt in relation to the interpretation of the graphs. >> >>[not normalized] >>http://www.lbm.fcav.unesp.br/download/todos.png >> >>[normalized] >>http://www.lbm.fcav.unesp.br/download/todos.norm.png >> >>They are: >> >>1. it has outliers. The presented amount is very much or is relatively >>normal for this >>platform? >> >>2. the box is above of line zero (y axis). That is normal? After >>normalized the data, they are between position 2 and the 3 (y axis). >>This can happen? >> >>We intend to pick up learn with yours commentaries, therefore, you are >>free to send us any others commentary about the data or quality check on >>this platform. Yours suggestions is very apreciated. >> >>Thanks very much >> >>Marcelo >> >>_______________________________________________ >>Bioconductor mailing list >>Bioconductor@stat.math.ethz.ch >>https://stat.ethz.ch/mailman/listinfo/bioconductor >> > > > > -- Best regards Wolfgang ------------------------------------- Wolfgang Huber European Bioinformatics Institute European Molecular Biology Laboratory Cambridge CB10 1SD England Phone: +44 1223 494642 Fax: +44 1223 494486 Http: www.ebi.ac.uk/huber
ADD REPLY
0
Entering edit mode
@sean-davis-490
Last seen 6 weeks ago
United States
On May 12, 2005, at 3:53 PM, Marcelo Luiz de Laia wrote: > Dear All, > > We work with cDNA in nylon membranes. We use bioconductor to analyze > our data, basically, the packages limma, affy and vsn. The > normalizations are made using normalize.quantiles from affy package. > Normalize.quantiles does not "center" the data around 0. > 1. it has outliers. The presented amount is very much or is relatively > normal for this > platform? This isn't the way to look for outliers. You might want to make a scatter plot of replicates against each other to see how closely they agree. > 2. the box is above of line zero (y axis). That is normal? After > normalized the data, they are between position 2 and the 3 (y axis). > This can happen? > As I mentioned above, normalize.quantiles will not center the data. You don't mention if you are using ratios or intensities (presumably?). If you are using intensities or a common-reference design, having the data centered around 0 is not that important. Sean
ADD COMMENT

Login before adding your answer.

Traffic: 212 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