DNAprint Genomics (DNAG)

[worktoplay]

This is just too easy. I choose F.

F. Quote directly from one of the Lead Researchers

This brings me to my next question. With the 500K Array Set and this next-generation array, we now have tools to assess SNP genotypes, copy number variant profiles and expression patterns. These could be thought to represent three non-overlapping dimensions of genetic information. How do you envision such information being combined in genetic studies moving forward?

Scherer: Ultimately, it would be ideal if we had a single platform that allowed us to detect all of these different genomic profiles from DNA to RNA all at the same time in one experiment. I don’t think that exists.

The next-best scenario is to have at least two of them. We already do, of course. You can assess SNP genotypes and then CNVs on the Affymetrix 500K Array, as we did for the consortium study. Overlaying the gene expression data will also be crucial because we now know that both SNPs and CNVs can either directly or indirectly affect gene expression.

Gee, it doesn't sound as if this researcher is telling us that SNPs are a thing of the past. In fact what he says is that they "know" that SNPs can either directly or indirectly affect gene expression, and that the best scenario is to have a genotyping system that can look at both SNPs AND CNVs.

Secondly, it is clear from your posts over on RB that your understanding of CNVs is that they solely represent instances of multiple copies of genes or sequences. If you'd do a little reading prior to making these outlandish statements you'd understand that a CNV isn't necessarily a multiple copy type of variant.

In fact, CNVs are not a "new" discovery. Geneticists have known about and studied them for years. The significance of the latest work is the size and volume that were discovered. But, there are many types of variations that are classified as CNVs. That 12% variation they refer to includes insertions and, in some cases, deletions of entire sections of chromosomes that had NO apparent affect on the individual. How much of the newly discovered variation has any significance at all is not known at this time. The conclusions contained in your posts have clearly NOT been drawn by the researchers that conducted the work.

They also know that there are more to find, but there are technical issues that need to be overcome. Most notably to THIS discussion would be the fact that in order to detect and study those in the more complex regions of the genome, they'll FIRST need to find MORE SNPs:

Because they were complex, they were often dropped out of the HapMap Project. So there may be a paucity of SNPs there. As a result of that, they are not well represented on SNP arrays, as one would expect.

So there is some ascertainment bias away from us being able to detect these regions and make correlation studies. If nothing else, we know now that these regions are complex. Even if we can annotate those that are simpler versus those that are complex, we still have to make a special effort to have a representation on arrays that are developed going forward.

For the bi-allelic-type CNV regions, we may be able to tag them using either existing or new SNPs in a typical manner. Based on the data in the Nature paper, we may be able to follow a CNV using a tagged approach about 25 to 30 percent of the time. But certainly for at least 50 percent of them, either new representative SNPs will have to be developed. Alternatively, you might develop a tailored array that would cover SNPs and CNVs using either classical SNP coverage or non-SNP oligonucleotides targeted to specific regions of the genome. That would be very, very valuable.

Are we getting the picture here Frog? The SNPs provide the signposts on the roadmap around the genome. The SNPs enable researchers to delineate a region so that they CAN study it.

BTW, Frog, I thought you told TonyTox on the other board that you couldn't do this work using existing equipment. In fact I'm certain you did. Here's the text:

http://ragingbull.quote.com/mboard/boards.cgi?board=DNAG&read=362547

...and what about the equipment manufacturers? do researcheres still use genotyping machines for this work or do they use something else?

They certainly can't use them in their existing state. They will either need new machines or more complex software.

So how did they manage to do this on an Affymetrix 500K Mapping Array? In fact, Affymetrix was touting this accomplishment quite proudly. You said it couldn't be done. I guess they didn't check with you first, huh?

I could go into some of the other erroneous statements contained in that TonyTox response, but we'd be here all day.

The bottom line is that you've never really understood DNAPrint's classifiers (hell, I don't think you even understand that it's not necessarily the SNP's themselves that cause the variable response.) The SNPs are markers, signposts that point to relevant sections of DNA. Many times NOT linked genetically to the trait that the classifer infers. IMO, far from making the classifiers obsolete, this new information could prove quite valuable to the company in helping them better understand the biology behind the inferences provided by the classifiers.

How does a seemingly unrelated region of DNA impact a trait without being genetically linked to the trait? Maybe CNVs will help explain. Equally possible, maybe they can't.

Later,
W2P