Question: SCAN UPC normalisation
0
gravatar for sanj
4.7 years ago by
sanj20
United Kingdom
sanj20 wrote:

Hi,

 

I am using SCAN.UPC package to get batch effect adjusted intensities for two HGU133Plus2 datasets. However for few probesets I get negative intensity values. I understand that the data is log transformed but is there any way to avoid  that and get the intensities on a linear scale without any negative values?

Also,is it possible to use SCAN.UPC with the Illumina-HT12V4 bead chips?

Thanks.

 

normalization scan.upc • 826 views
ADD COMMENTlink modified 4.6 years ago • written 4.7 years ago by sanj20
Answer: SCAN UPC normalisation
1
gravatar for Stephen Piccolo
4.6 years ago by
United States
Stephen Piccolo560 wrote:

This should answer your first question: Expression Values for SCAN.UPC?

I'm looking into the possibility of adding support for Illumina and Agilent single-channel arrays, but currently these are not supported. However, you can apply the UPC method to these array types if they have been preprocessed using other methods.

ADD COMMENTlink written 4.6 years ago by Stephen Piccolo560
Answer: SCAN UPC normalisation
0
gravatar for sanj
4.6 years ago by
sanj20
United Kingdom
sanj20 wrote:

Thanks for the reply.

After the SCAN normalisation I have used the UPC detection method on around 4 different datasets, 3 of which are on affymetrix platform and 1 is on Illumina HT-12 V3 expression beadchip. The distribution of the detection value seems to be on extreme ends that is either 0 or 1 and very few probesets have detection value in the range of 0.1 to 0.9 irrespective of the dataset or the platform used. Can someone help me understand why is this the case? In the UPC paper it is mentioned that anything above 0.5 is considered as detected by the authors, but looking at the graph  0.5 does not look very different then 0.4 or 0.6 cut-off.

 

 

 

 

 

 

ADD COMMENTlink modified 4.6 years ago • written 4.6 years ago by sanj20

The idea behind the UPC method is that some genes will be "on" while other genes will be "off." They are probabilistic values, so you can get a sense for how much uncertainty there is in estimating whether a given gene is on or off. The 0.5 threshold is definitely not a hard and fast rule. It is more of a guideline. As you mentioned, it may not be straightforward to say that a gene with a UPC of 0.4 is clearly off whereas a gene with 0.6 is clearly on. To get a better idea of the expression levels, it is often helpful to look also at the actual expression values.

ADD REPLYlink written 4.6 years ago by Stephen Piccolo560
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: 201 users visited in the last hour