Decode your dna for $1000 (1 Viewer)

[pete]

http://www.wired.com/medtech/genetics/magazine/15-12/ff_genomics

It takes about 10 minutes of slavering to fill the 2.5-milliliter vial that comes in the fancy lime box provided by 23andMe. Wrap it up, call FedEx, and two to four weeks later you get an email inviting you to log in and review your results. There are three main sections to the Web site: Genome Labs, where users can navigate through the raw catalog of their 23 pairs of chromosomes; Gene Journals, where the company correlates your genome with current research on a dozen or so diseases and conditions, from type 2 diabetes to Crohn's disease; and Ancestry, where customers can reach back through their DNA and discover their lineage, as well as explore their relationships with ethnic groups around the world. Family members can share profiles, trace the origin of particular traits, and compare one cousin's genome to another in a fascinating display of DNA networking.

https://www.23andme.com/

 

[Jill Hives]

I am trying to decide if that is very cool or scary.

 

[Denny Oubidoux]

i wouldnt mind getting a test out of curiosity but not by googles missus

 

[flashback]

A bit light on details.
I guess they are using Solexa. Those machines are fierce finicky.

Anyway, knowing what your dna is, and knowing what your dna does are very far apart.


They have had humans sequenced, and assembled for a while, but they still dont know how many genes* are in a human. I am talking about orders of magnitude here.

I dont like the word gene. Its sort of a pointless word. It means much the same as locus now. Talking about transcripts, or Coding sequences makes sense. A gene is just really a point on a genome now. (Which is the same definition as a locus.)

And they have no idea how many transcripts are encoded on the human genome.

So, basically, all this "cracking" the genome talk is bullshit. They have sequenced it, they have assembled it. They have a bit of a vague idea about what is going on on it. Cracking implies they suddenly unlocked some code, which is just not the case. Or, is no more the case now than it was 60 years ago.


Just for the record, another bullshit phrase is "the cure for cancer". Like there is a cure. A single cure. There are more cancers conceivable than there are tissues. One thing is not ever going to cure them.


Sorry about that.

 

[pete]

 

[flashback]

And that 23AndMe site is pretty funny.
I am glad someone has it all worked out.
Their Genome Explorer looks a bit shitly.

I mean all they do, is get your sample, send it to Illumina for Solexa sequencing, do a SNP Chip and thats it. They dont actually sequence your genome. They just run that SNP chip.




And trying to figure out everything about you using just a SNP data is weak. To say the least.


Presumably it will make a fortune.

 

[nEiLo]

And that 23AndMe site is pretty funny.
I am glad someone has it all worked out.
Their Genome Explorer looks a bit shitly.

I mean all they do, is get your sample, send it to Illumina for Solexa sequencing, do a SNP Chip and thats it. They dont actually sequence your genome. They just run that SNP chip.




And trying to figure out everything about you using just a SNP data is weak. To say the least.


Presumably it will make a fortune.

i'm guessing you know a bit about this?

 

[flashback]

i'm guessing you know a bit about this?

Judging by the state of the shite I am coding up today, I am starting to wonder.

 

[snakybus]

do you work in bioinformatics flashback, yeah?

what's that like?

 

[flashback]

I do yeah.
What's it like? Hmm. I have to say its hard.
That would be my first answer.

Like, to be a bioinformatician, (I think) you have to :

be very strong in a computer language, Perl, Java, Python or maybe Ruby. Maybe some others. I dunno, but you have to be able to code effortlessly, as in not be thinking about the language when you are coding. Also be able to do some software engineering.

have a decent ability on command line UNIX/Linux tools, and understand a bit how Linux works, how to get what you need to know out of massive log files and things.

have some sort of mathematical cop on. Nothing major. You would to have some 3rd level maths, or be able to understand it at least, not be freaked out by integral signs and stuff.

understand some genetics, and some biology.
I used to say that biology is the least important part, but, actually you do need to have a fairly good idea of how things, specifically genetic things, work... otherwise you will be making fuck ups all the time.


So, computational biology is weird. You dont have to be an expert in any of the above, but you have to know your shit in all of the above.

Having said all that, I really only scrape through on the above, and I find it hard.
You are also working with smart people who are often experts in one of the above, and who dont really appreciate how hard or easy certain things are.

Ie, what happened over the last few months, I spun up a prototype app fast, everyone is amazed.

I turn the prototype into something that can pipelining 100s of thousands of experiments rather than a couple of hundred, and basically nothing else, and people are pissed because I am "not doing anything", or I am suddenly "working really slowly". And then there ended up being a massive fight, and people are not talking any more. (Even though the work was reasonable enough to be published.)

The numbers you are dealing with almost always huge, and error checking can be really hard.
aaatttgccttt might be the right answer, but aaaggcaaattt might not be. Or they both might be the right answer, because aaatttgccttt is the reverse compliment of aaaggcaaattt. (DNA runs in both directions. Which means its a head fuck.) And, when you are dealing with Genomes, which are big, errors will be found. I dont think I ever wrote something right the first time around. Ever. You have to assume you are wrong, somehow, and figure out how.



This answer is far too long.

Anyway, I find it hard, but I am not the smartest, or most focused person in the room. The money isn't the greatest, but it could be worse.

I haven't been fired.
That's another plus.

 

[Lili Marlene]

bet Dawkins can't do any of that

 

[Scientician 0.8]

They sell your info to pharmaceutical companies etc. don't they?
I suppose it would be really good for someone who was adopted and wanted to know their genealogy precisely.

 

[snakybus]

bioinformatics

Sounds prety cool. I wrote an piece on it a couple of years ago and it seems amazing. The potential it has seems limitless.

 

[flashback]

Sounds prety cool. I wrote an piece on it a couple of years ago and it seems amazing. The potential it has seems limitless.

Like anything new, and shiny, there is a huge amount of hype around it.
A lot of people dont understand computers, and dont understand genetics.

There is this idea, propagated by programs like Heroes, and fear mongerers, that suddenly every thing is changed now that we have sequenced and assembled various genomes.

Bioinformatics is a tool. It is powerful, and it is exciting, but it is just a tool that allows us to look at organisms.

The major point is that nature imposes some hard limits on what is possible. So, where you will have companies that can sequence genomes and flag people for gene therapy who are prone to some cancer or whatever, this doesn't mean all bets are off on what can be achieved by animals.

Its hard to explain my point here.
Nature is the most parallel, most powerful test that can be conceived of. Evolution is the most powerful solution that can be conceived of.
Organisms are being pushed, and have been being pushed as hard as they possibly can, for longer than humans can imagine.
So, the solutions are pretty good.

The idea that humans, with their little computers are going to come along and not only figure out everything, but improve upon and perfect it is not going to happen any time soon.
Things like gene therapy, designing better drugs by understanding protein folding, things like that will happen. But all of these must act within the limits imposed by nature.
All bioinformatics can do is make suggestions to experimentalists, who must go off to their systems, and test this suggestion within their system's bounds.

Bioinformatics is a deadly Makita power drill, that allows all the lads on the site to throw together things that they used to have to spend ages doing. Its such a good drill that maybe different concepts can be employed on the building site.
But, the building site is still exactly the same. Gravity still pulls things down. Newton's laws still rule.

 

[sarah]

nice answer.