Hi everybody,
I want to analysis gene expression in GEO2R from NCBI, but GEO2R just analyze dataset that have biological replicates (pair data). of course, I have experience in R for affymatrix data using bioconductor from this case study (http://bioinformatics.knowledgeblog.org/2011/06/20/analysing-microarray-data-in-bioconductor/) that is CEL format but I need other scripts in R to do other format (txt data) for other kind of platforms of microarray like 3D oligo chip thus I try this ([https://www.biostars.org/p/148471/) but it is not worked. So, main question is, can someone guide me how I can proceed with limma or other packages in R, if I have data that is txt format and do not affymatrix with no biological replicates (for example GSE56875) in such format? I think if we can do some alteration in GEO2R script to do single datasets, it would be success
Many Thanks in advance
0 answers
No answers yet.
Log in to answer this question.
-
Tutorial for analysing gpr files of microarray
written by afli
-
How can I loading and normalising the data for (HuGene-1_0-st-v1) microarray Data In Bioconductor
written by AAhmed •
-
capturing phenotypic information in R
written by 12021560-002 •
-
cdf environment for cytoscan hd array
written by arti.pandey69 •
-
Differential expression analysis using Limma (Not on cel files)
written by Bioinformatist Newbie
-
Finding differentially expressed genes in Control vs. Treated
written by Bioinformatist Newbie
-
Creating a log intensity plot of normalized microarrays
written by StatsBio •
-
When Should You Do Spatial Normalization On A Single-Channel Microarray And Which R Package Should …
written by scottwilliamrobinson
-
Puzzling Error Message While Working Through A Bioconductor Tutorial On Microarrays
written by Mycroft34 •
-
Tutorial: Analysing Microarray Data In Bioconductor
written by Obi Griffith
Can you try this, perhaps (assuming you're starting with CEL files): A: HuGene-1_1-st accessory files for microarray analysis
Otherwise, you can obtain the normalised data with GEOquery by following the code here: https://www.ncbi.nlm.nih.gov/geo/geo2r/?acc=GSE56875 (see the R script tab).
Thanks Kevin, but this is not a good way for me. as a matter of fact, GSE56875 is not a CEL file that I do it in ([ A: HuGene-1_1-st accessory files for microarray analysis])[1] beacause it is for CEL data not txt. and about (https://www.ncbi.nlm.nih.gov/geo/geo2r/?acc=GSE56875) I should say, in this data we have 10 data set that only one of them (D7-GMT-miR-133-iCM) with accession number GSM1370504 and GSM1370507 are biological replicate that we can analyzed them in typical GEO2R as a treatment with MEF (GSM1370499) as a control. but, for other sets of this data, we don't have a biological replicate. So, the problem is that the R script for this kind of data not work, thus, we need R script that work in this situation for single data with no replicate as a control or treatment.
If I have understood you correctly, you want to make a few comparisons consisting of just 1 sample versus another - is that correct? That is not an ideal situation and one would likely make a high percentage of false-positive statistical associations that way.
Dear Kevin, of course, your answer is correct but it is an only way to do this kind of data. do you have any plan for this situation?
Hello, but, when you think about it, how can you faithfully compare 1 versus 1? Each gene will be just 1 value versus the other, like, 2.5 versus 2.6? The best that I would advise is to just derive a ratio for each gene (2.5 / 2.6 = 0.96). Even then, I think that you will struggle to find credible results.
Hello, maybe it is a good way. but what value do I compere? in NCBI, on the one hand, we have data of GSE56875 in this page ( [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE56875])[1] that consist of series matrix file, SOFT and MINiML. Series matrix file including Mean and median of each spot (show one gene) that I think it is maybe good for doing our job in something like SPSS but i am not sure. on the other hand we have a raw data of each sample including normalized signal intensity, for example in this two page that I want to compere (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSM1370499)][1] MEF as a control and ([https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSM1370505])[2] D18-GMT-iCMs as a tretment. My exact question is that is it a good way to derive a ratio of treatment /control base on normalized signal intensity to achieve to gene expresion? is it a normalize data? or we need to first normalize data and then use the treatment/control ratio? maybe we have other plan, can we make a copy from single file to create duplicate file and then do R script like GEO2R with some modification to do this way?
You may want to follow this recent thread, where the user also wanted to compare 1 sample versus 1 other (cell-lines); however, it was RNA-seq: C: Differential analysis between two cell lines
In your case, I would normalise all samples together and then conduct the statistics as usual (with limma). However, the reproducibility of the results, considering extremely low sample numbers / replicates, is a major issue.
Many Thanks Kevin for your time and answers, but RNA-Seq is completely different from microarray analysis. moreover this data is normalized I think,and then use edgeR from R. I suppose if we do some changes in ([https://www.biostars.org/p/148471/) it would be worked for our analysis. But I am beginner in R and I could not to do this script in R.