This is a test version of Biostars. For the public version, visit https://www.biostars.org.
illumina gene expression

Hello all

i am dealing with certain illumina microarray data.

i working on genomestudio and proceeded the way it is given in user guide. But i am facing problem, while loading files (.idat) in repository tab, once i click on folder appearing(Barcode) on sentrix array it is not recognizing the sample (idat files). So what all files do require to keep in folder where my idat files are saved? and what is the reason that it is not recognizing my files.

genomestudio microarray geneexpression illumina
• 6,632 views • link
updated by Ram Ram • written by Kritika Kritika

Actually just going through manual of genome studio gene expression I kept idat file only in one folder. But after that I kept all files in same folder so it works. I will try bioconductor also for my data.

Anyways thanks andrew and poisonalien for helping me.

• 0 views • link
updated by Ram Ram • written by Kritika Kritika

1 answer

Not sure about GenomeStudio. But if you are comfortable using R, use this script. It takes idat files as input, does normalization and performs differential expression between two groups. (Assuming there are no batch effects)

Usage:

source("AnalyzeBead.R")
result = beadAnalyze(idats = c("file1.idat","file2.idat","file3.idat","file4.idat"),names = c("control1","control2","treated1","treated2"),condition = c("control","control","treated","treated"),ref.condition = "treated")
• 0 views • link
updated by Ram Ram • written by poisonAlien poisonAlien

Just to tack onto poisonAlien's answer, the behaviour you're seeing in GenomeStudio is just a quirk of their software, and I'm sure there was a reason for it once upon a time, you need the IDATs to be in a folder separated by chip ID (each folder is the chip ID number), in addition you'll need SDF files in the folder too. GenomeStudio is not as flexible as bioconductor methods for analysing microarray data, so I'd second poisonAlien's answer, to try the analysis in R, you'll get more of an appreciation for what actually occurs in a typical differential expression analysis. If you still have trouble with GenomeStudio, I'd suggest you contact Illumina support, you've paid for a license, so you should make use of the support they provide.

• 0 views • link
updated by Ram Ram • written by andrew.j.skelton73 andrew.j.skelton73

Hello poisonAlien

The script which you shown above is giving error:

Error in idatData$Quants[, "CodesBinData"] : subscript out of bounds

Please tell me what this error means and how to rectify it.

Thank you

• 0 views • link
updated by Ram Ram • written by Kritika Kritika

What platform are you using? chip ID?

• 0 views • link
updated by Ram Ram • written by poisonAlien poisonAlien

Currently I am working on one dummy sample.

• 0 views • link
updated by Ram Ram • written by Kritika Kritika

You gotta be more specific. That code assumes that you're working on human arrays (to be specific HT12 v4 chip, because that's what we use frequently in our lab). If you're using another array, you will need to change the annotation. Do you have replicates ? And do you have all the libraries installed? (beadarray, limma, illuminaHumanv4.db)

• 0 views • link
updated by Ram Ram • written by poisonAlien poisonAlien

Yes, the chip is HT12v4 I confirmed from where I got the samples. Yes, all libraries are installed

• 0 views • link
updated by Ram Ram • written by Kritika Kritika

Can you post your command?

• 0 views • link
updated by Ram Ram • written by poisonAlien poisonAlien 
source("Microarray/AnalyzeBead.R")

result = beadAnalyze(idats = c("/dummy_data/Image Data/9666412702/9666412702_A_Grn.idat" , "/dummy_data/Image Data/9666412702/9666412702_B_Grn.idat"),  names = c("control","treated1"), condition = c("control","control","treated","treated"), ref.condition = "treated")

 Error in `[<-.data.frame`(`*tmp*`, , "sampleFac", value = c("control",  :
  replacement has 4 rows, data has 2
• 0 views • link
updated by Ram Ram • written by Kritika Kritika

Ahh! See you are providing two idat files (one treated and one control) but your condition says two control and two treated. That's what your error report says.

Try:

result = beadAnalyze(idats = c("/dummy_data/Image Data/9666412702/9666412702_A_Grn.idat" , "/dummy_data/Image Data/9666412702/9666412702_B_Grn.idat"),  names = c("control","treated1"), condition = c("control","treated"), ref.condition = "treated")

Note, you don't have replicates so you wont get any p-values.

• 0 views • link
updated by Ram Ram • written by poisonAlien poisonAlien

Oh! Thank you :)

poisonAlien can you please explain me this line

names = c("control","treated1"), condition = c("control","treated"), ref.condition = "treated")

If I have replicates then what command should I use? Same as the above you provided

I tried to understand the source code of this but it's going out of my understanding

Thanks

• 0 views • link
updated by Ram Ram • written by Kritika Kritika

What is understood from this command is names=c("control","treated1) will refer to object of control and treated condition = ("control" , "treated") will handle error or warning?

What ref.condition this?

• 0 views • link
updated by Ram Ram • written by Kritika Kritika

According to this commands

result = beadAnalyze(idats = c("file1.idat","file2.idat","file3.idat","file4.idat"),names = c("control1","control2","treated1","treated2"),condition = c("control","control","treated","treated"),ref.condition = "treated")

file1.idat, file2.idat are replicates for treated and file3.idat file4.idat are replicates of control? Am I correct?

As I said already I m dealing with dummy data I tried some more sample so after running this command I got message :-

Annotating control probes using package illuminaHumanv4.db Version:1.26.0
Calculating array weights
Array weight

After typing

result

it is showing certain values with column

  ID                        logFC      AveExpr             t      P.Value adj.P.Val         B
 ILMN_XXXXX
• 0 views • link
updated by Ram Ram • written by Kritika Kritika
  • idats is vector of your dat files (in the above example there are 4 dat files)
  • names is sample names for those dat files (above they are named as control1, control2, treated1 and treated2). Yes, they're replicates.
  • condition is sample characteristics. First two are control and last two are treated. It can be anything based on your experiment. (like knockdown, over expression, etc.)
  • ref.condition is which one of the condition to use as a reference. Here I am comparing everything with treated. All up or down genes are with respect to treated samples.

Output is typical limma results. You may want to read limma manual. In short, logFC is fold change with respect to control samples, AveExpr is average expression across all your samples, t statistics, p-value, adj.P.Val is FDR, B is odd ratios. Also there are other stuffs like Probe sequence, probe quality, its locus on genome, where it lies on transcript, etc.

Script itself is well commented, so you should be able to follow. However life will be easier if you know how expressionset object is represented and its slots in Bioconductor. Tomorrow I will update the script with PCA, you can check again.

• 0 views • link
updated by Ram Ram • written by poisonAlien poisonAlien

Very useful information

Thanks a lot PoisonAlien

• 0 views • link
updated by Ram Ram • written by Kritika Kritika

Hi Kritika,

Same like yours I need to do gene expression analysis with IDAT files. Could you please tell me how did you do your analysis? Workflow and packages.

Thank you

• 1 views • link
written by Vasu Vasu

Hi poisonAlien,

I am trying to use your Source script (AnalyzeBead.R), but am running to this error. What could be the problem?

result = beadAnalyze(idats = c("4487653088_J_Grn.idat","4487653088_K_Grn.idat","4487653088_L_Grn.idat","4487653151_A_Grn.idat"),
                      names = c("4487653088_J","4487653088_K","4487653088_L","4487653151_A"),
                    condition = c("day0","day0","day2","day2"),
                      ref.condition = "day0", fdr = 0.05, plotPCA = T)

Annotating control probes using package illuminaHumanv3.db Version:1.26.0
Calculating array weights
Array weights

Error in `levels<-`(`*tmp*`, value = if (nl == nL) as.character(labels) else paste0(labels,  : 
  factor level [4] is duplicated

Hope to hear from you.

Cheers

• 1 views • link
updated by Ram Ram • written by evanskataka • evanskataka

Please use ADD COMMENT/ADD REPLY when responding to existing posts to keep threads logically organized.

This should be posted as a comment under poisonAlien answer.

• 1 views • link
written by GenoMax GenoMax

Log in to answer this question.

More posts like this