What does DNA mean to you? #14

by Dr. Hsien-Hsien Lei
Posted July 16, 2008 in Meaning of DNA

Andrew Yates of Think Gene is feeling blunt today as he tells us what DNA means to him.

Nothing.

My background is computer science, so to me, DNA is the object code of life. Unlike human-designed languages, DNA is entirely unbounded by intelligibility or elegance —only function.

So we are looking for meaning at the wrong level of abstraction. Our understanding of DNA is tainted by an anthropomorphic misunderstanding of how a language “should” work: “genes” are “sequences of letters” positioned by an “index” like words in a book. One word, one meaning[1]. This is a mistake, as supported by the inconsistent success of genome-wide association studies (GWAS) and the disappointing usefulness of today’s genomic testing.

We don’t try to understand object code in software without abstraction, so why do we try to understand DNA directly in life? Here’s why we shouldn’t try to understand DNA directly —even more so than object code:

* DNA is an implementation, not a map of abstractions. That is, units of DNA have no constraint to “mean” anything. Even object code can usually be interpreted as processor instructions and numbers.
* DNA is a template for amino acids and RNA, not a set of instructions (code) or table of facts (data).
* What DNA “describes” is probabilistic, dynamic, highly context-sensitive. It moves. Its parts move. Its environment moves. It’s chemistry. Object code is discrete and static. It’s math.
* DNA is hard to sequence. Object code is trivial to sequence.

Genomics today is like alchemy: we’re tinkering with a system we don’t understand in hopes of some elixir of longevity —except we call it “the cure for cancer.”

Why? Because we are impatient. Because we vastly over-estimate our ability to understand complex systems without simple abstractions. Because we believe what is difficult must be valuable. Because genomic research today is commercial, and gold must be made.

Well, that’s crap.

In software, we abstract object code with higher-level languages. When that system becomes too complex, we make a new, even higher level interface and abstract again. We continue until surface complexity is low enough to be useful.

In genomics, we label genes with some incomprehensible, ontologically-inconsistent name and then strain to make that gene “mean” some attribute or disease.

There is some use for the black-box, top-down genomic testing, but I believe that this approach alone is wrong. I believe that what we should be doing is creating better abstractions, interfaces, by which DNA can be understood. I believe that the future of genomics —the people who will make DNA mean something— will be the language designers who compile to DNA.

Until then, God laughs. There’s a reason why Window’s object code is everywhere, but the source code is top secret. Bill Gates laughs, too.

RSS feed | Trackback URI

Please note that comments left using the form below will be publicly displayed. If you'd like to correspond with me privately, please email me at hsien@eyeondna.com.

If your comment doesn't show up immediately, it's probably in moderation. I will approve it as soon as I can! Thanks for your patience.

4 Comments »

Comment by David Bradley

2008-07-17 12:44:25

Best one yet Hsien!

Hope you’re latest bundles of self-replicating DNA is thriving

db

Reply to this comment

 

Comment by Celena Mabry

2008-07-18 21:43:37

DNA is a funny thing. I have an aunt that was adopted when she was a baby and you would think being raised by my grandparents that she would be just like the ‘natural’ children but she isn’t. We found her mother years later and realize just how much DNA played in who she is and not who raised her. Amazing stuff.

Thanks,
Celena

Reply to this comment

 

Comment by Andrew Yates

2008-07-21 09:12:38

Aw, “dna dundee Andrew Yates” was edited out? I like that.

Reply to this comment

 

Comment by Jean Vincent

2008-09-30 00:08:09

Agree on the object code there is no such thing as source code on top of DNA.

All we have is the user interface (psychology, symptoms checking) to study at the top and the chemistry at the bottom.

We can also use the debugger (neuroscience, surgery, imaging) from time to time to provide some insights.

We both need the top-down and bottom-up approach to understanding life and of course a bit of patience.

Understanding life will indeed be much more complex than we would like it to be. But this is what makes science interesting after all.

Reply to this comment