Hello,
I am looking for a software allowing the user to compare the results of microarray experiments, but taking into account different platforms. I mean to compare results from i.e. Agilent and Affymetrix arrays.
I have found MicroarrayRUs, but it supports only Affymetrix, and Illumina.
I will be grateful for any suggestions.
What exactly do you mean by comparing "results" from different platforms? It is not a simple thing to do and I don't know of any software that will simply allow you to feed in raw CEL files and Agilent TXT scan files and output a comparison.
For one thing, they use completely different probe sets, so mapping probes to genes is one issue. If you actually want to map probes to probes, that could be done as well, for those that interrogate identical genomic transcripts, but again would be a bit of work as their would be many ambiguities to deal with and I don't know of any existing software tool to do that automatically for you.
Data normalization is another issue, as methods suitable for one are either not suitable for the other, or at the least, are not optimal for both. If you independently normalized each dataset, you could potentially merge them with an additional normalization or scaling correction, but that also can be very problematic (for one thing, you risk compressing the data so much that you lose all ability to discriminate any differences).
Most such cross platform comparisons are done by analyzing each platform data set independently using the optimal normalization algorithm and statistical test procedure for each (but usually with a common criteria of significance such as FDR2), then comparing the resultant lists of significantly differentiated genes. If you wanted to keep the statistical analysis identical, you could do that on the normalized signal values, using for example a standard ANOVA with each data set, or LIMMA and eBayes in R&Bioconductor. But there is no single normalization approach that would be optimal for all platforms - the platforms are simple not comparable technologies in that regard.
Once you have independently analyzed each data set, you can compare directly the overlap in significantly differentially expressed genes, or you can compare the two lists based on what functional ontology signal they appear to have captured, as two different gene lists may still contain elements for the same functional ontology categories, as determined by traditional enrichment, GSEA or Connectivity Map analysis or some other functional biology comparison. See for example the 2006 paper by Cheadle et al. comparing Affy, Agilent and Illumina array data sets - http://www.sciencedirect.com/science/article/pii/S0890850806000545
Comparison of different (microarray) platforms is a goal of the MAQC/SEQC consortium:
http://www.fda.gov/ScienceResearch/BioinformaticsTools/MicroarrayQualityControlProject/
In 2006 a special issue of Nature Biotechnology was released containing set of publications dedicated to this topic:
http://www.nature.com/nbt/focus/maqc/index.html
You will not find there any software but rather description of metrics which could be used for a such comparison (more less as described by Michael).
Extension of this to RNA-Seq has just been published by the way:
http://www.nature.com/nbt/journal/v32/n9/index.html
Although combining results from multiple micro-array experiments provide greater statistical power to your final results, the task is quite difficult. There is no normalization method that works efficiently for rawdata from affymetrix, illumina, agilent, codelink and spotted oligo-nucleotide arrays. However, we made an attempt in this regard recently.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3091663/
The focus has been more about processing data separately and deriving consensus across experiments. I believe that approach would definitely help you in obtaining most reliable result.
To know about more meta-analysis methods please refer to this review
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3898528/
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Software