Hello all,
Can anyone help me with the following problem I’m facing?

For my project, I need to integrate common information (genes) of 4 platforms for the purpose of gene clustering analysis.

I can’t do cross-platform integration of data directly as:

I understand that it is bit tedious to do so as they all have used different total RNA levels, and one of them has a different pre-processing (normalization) technique. I also feel that scanning resolution might be an important factor to consider.

The few commonalities I observed are:
I have taken all the data from Gene Expression Omnibus (GEO) at NCBI.
All series in the platforms have dual-color experiments (Cy3, Cy5)
All the values across series in the arrays are normalized log2 ratios(Cy5/Cy3) or treated/ control ratios.
All series are “total RNA” for both channels.
All series have 2 or 3 replicates.
All the data is from time-course studies.

The differences I observed are:
I’ve considered the following platforms:
1. Synechocysis sp. PCC6803 full genome array version 1 (GPL965)
2. Synechocystis sp. PCC 6803 oligo array (GPL7016)
3. Takara CyanoChip v.2.0 or IntelliGene® Cyano CHIP Ver.2.0(GPL3214)
4. Takara CyanoChip v.1.2 or IntelliGene® Cyano CHIP Ver.1.2 (GPL3095)

The above have been normalized respectively as follows:
1. Global normalization
2. Loess Smoothing, ANOVA linear model
3. LOWESS from GeneSpring
4. LOWESS from GeneSpring

The amount of total RNA used are respectively as follows:
1. 16 microgram
2. 8-15 microgram
3. 10 microgram
4. 25 microgram

Scanning methods used respectively:
1. GenePix Pro 4.1
2. Axon GenePix 4000B laser scanner (resolution of 10 µm per pixel)
3. GenePix 4000B Scanner at photomultiplier tube settings that afforded the best overall balance between Cy3 and Cy5 channels.
4. GenePix 4000B fluorescent scanner and image intensity data were extracted and analyzed with GenePix Pro 4.0 analysis software.

Kindly suggest me a way that I can integrate the common genes featured in these arrays for the future purpose of clustering.

Do I have to treat replicates first and then carry out integration or vice-versa?

Do I fill in missing values before or after integration of data?

How can I make sure that the values after integration are comparable across the arrays for the genes?

Please do reply here or at: rnayana6@gmail.com
I’d be very much obliged.

Thanks and best regards,
NAYANA RAMACHANDRAN
MSc Bioinformatics,
University of Pune,
Pune, India.