Friday, December 19, 2008

Ouch!! CNV with lackluster results....


All it takes is 2 seconds to step on some of my readership's toes and I feel it. Yesterday I posted on a 5% error rate for Whole Genome sequencing, I argued that even at 30x coverage it would not be ready for clinical diagnosis. I had CEOs of sequencing companies emailing me and VPs calling me. I even had pound for pound one of the best bloggers in the space say he was embarrassed for me.....Ouch!
Why do I get pushback from people, when all I am doing is throwing some cold water on the party???

Get ready, because I am about to throw some more.....Remember yesterday when I said SNPs were one of 7 or 8 factors that will differentiate each of us??? Well, CNVs are another of those 7 or 8, 2 more include histone modification and methylation, telomerase activity and size would be another factor, the rest I am saving for my own....for now. I first heard about CNV is 2006 when Mike Murray at Harvard keyed me into these guys, since then I have been following the literature and hoping we could get some results....well, we have but......

Here's the cold water, CNVs are not everything either, despite what some very learned people say.....just like genetic and genomic testing is only PART of the armamentarium for a personalized medicine specialist, CNVs are only part of the story.
True, we may find some very high Odds Ratios and some very specific diagnostics in the CNV space.....unfortunately, the American Journal of Human Genetics lays an egg with a chinese study of osteoporosis CNVs that lead to an Odds Ratio of 1.7 for osteoporotic hip fracture. I was hoping some of these CNV stories would be much more exciting than SNPs....My guess is that alot of the SNPs that we found previously with GWAS may actually just be markers for CNVs....and if that is the case, can we expect that much more from CNV than SNP?


I know some who would say yes, and I look forward to their comments. I think we may see this as the key in some areas, where amount of transcript plays a huge role, like metabolism of compounds or perhaps in cell signalling and migration events, but what about diseases that don't need that so much, structural protein diseases, ciliopathies, etc......

Most importantly, what about the diseases that sneak up on us over time like diabetes or atherosclerosis? I don't think that these will be the answer here. I have a very strong feeling that my equation Genome + Environment = Phenome + Metabolome will still hold true....

The one thing I am certain of is how to make things clinically applicable and right now, CNVs, SNPs, or Whole Genome Scans....there are only a very few limited cases where we can use this stuff......Until the sequencing companies are willing to take the liability for how their product are used, there are going to be problems trying to sell it as medicine without medical professionals......
So sorry to GC, PM, CV, JR, DM and who ever else decided to email me or call me expressing their problems with my cold shower: shake it off, look for solutions and get back to climbing the mountain.


The Sherpa Says: I just want to keep the marketers from overhyping.....Because if we don't, they will "create" our science.....Through slick words and number play published in the New York Times or Wall Street Journal.

5 comments:

Andrew said...

Dude, people are pissed because you were sloppy, not because you were right and they were wrong.

Granted, if it was attention you wanted, it's attention you got.

I don't know, I thought it was obvious that this stuff wasn't ready for medical application... until the company published "use this in medicine and yes we're liable for saying that."

Is that a minority opinion or something? Am I just crazy or is everybody else?

Daniel said...

Why do I get pushback from people, when all I am doing is throwing some cold water on the party???

Wow, Steve, I'm impressed - rather than take the easy road of simply admitting you were wrong, you've bravely undermined your credibility even further! How does it feel? Liberating?

Two hints for future reference: (1) trying to disguise simple ignorance as bold iconoclasm is unlikely to be appreciated by people who actually know what they're talking about; and (2) the best way to "keep marketers from overhyping" is to combat them with accurate information, not to just post whatever comes off the top of your head.

As for the rest of your post... you know what, I just can't be bothered.

Steve Murphy MD said...

Daniel,
I recently attended an excellent lecture and discussion with Allen Bale, precisely about these sequencing techniques, I rotated through labs which have used them and I am familiar with the technology.

5% Total error rate is what it says it is....

Forgive me for being blunt, but your PhD better than MD mentality is juvenile and frankly offensive.

I know what you are doing and who you are speaking with. You seem like a nice guy, until you hear some of the things you have been saying about me.

So before you go too far....just stop.

5% total error rate is what has been quoted. I have spoken with Patrice Milos directly about this....it can be overcome, but 100x coverage is what is needed.

If you are looking to get into a petty schoolyard fight, then just know, I am way to busy to be involved in such foolishness, even with your antagonism....

I know more about medicine, you know more about science....Happy now?

-Steve

Daniel said...

Hi Steve,

your PhD better than MD mentality is juvenile and frankly offensive

I genuinely have no idea where you got this from. Of course I think that both scientists and clinicians are important in this game. In fact, back when the whole genetics blogosphere was enraged about the California debacle, I was one of the few scientist bloggers pointing out that clinicians know lots of stuff that we can't just learn from Google.

As for the rest of your comment - check your email. I've made my point, and I have no intention of turning this into a public brawl.

Anonymous said...

Man, this was just getting interesting. Just kidding.