Hacker News Reader

Comment by fragsworth | Hacker News Reader

Likelihood of cancer may be related to the general chance of reproductive mutation, and the chance of reproductive mutation is also likely to be something that changes through evolution.

If you have a species in a high-radiation environment, is it not possible that it evolves to have fewer reproductive mutations in general - and therefore lower chance of cancer?

View on HN

Yes, I think it makes sense that certain arrangements of DNA would be less susceptible to going cancerous than others. In one strand you might have a possible '1 bit flip' (I'm not a biologist in the slightest) change that would turn the routine into "while(1) {fork();}", but perhaps in another functionally equivalent strand you would need to flip 2 or 3 bits to get the same effect.

So basically you want to select for DNA that has a large (DNA equivalent of) Levenshtein distance to cancerous DNA. Perhaps that is something that you could even deliberately engineer (though I don't think a general solution is possible since determining if DNA is cancerous might be the halting problem?)

Maybe it would be easier to construct a strand of DNA that has the same effect as another strand of DNA, but with more error-checking. Throw in a bunch of asserts or something to send cells with DNA that fails the error checking into suicide mode...

In practice, I suspect that we have numerous ways of dealing with mutations. However, if I recall my high-school biology correctly, during DNA replication there is error correction. My guess would be that evolution would tend to regulate mutations rates through the efficiency of error correction mechanisms (and the accuracy of the primary duplication mechanism) more so than a meta-evolution in the structure of DNA. Having said that, it is reasonable that there would be evolutionary pressure to have DNA that requires more mutations before becoming dangerous.

Responding to the content of your post. DNA is not executable code as much as data. The 'primary' purpose of DNA is to provide the blueprints for proteins. DNA sequences can be thought of in by grouping the base pairs in 3 called codons (this grouping is not reflected in the molecular structure). The beginning and end of a gene are marked by "start" and "stop" codons, which are just a specific DNA subsequence. Conceptually, a gene is converted into a protein by replacing each codon with its corresponding ammino acid (based on a lookup table) [1]. The function of a gene is determined by the chemical properties of the resulting protein.

As it turns out, there are parts of DNA that do not code for proteins, yet still have significant effect. However, I am not aware of any evidence of a naturally occurring use of DNA that resembles executable code.

[1]http://www.biogem.org/codon.jpg

I suppose what I am suggesting is something along the lines of littering the DNA with lots more "stuff that does nothing" that has a very small edit distance from "encodes the creation of a protein that kills the cell."

So while you might normally have:

  ... ACGATTACGATACG ....
  ---|--------------|---
       |
       useful protein

You would now have:

  ... ACGATTACGATACG TACC TACC TACC TACC TACC TACC TACC TACC TACC ...
  ---|--------------|----|----|----|----|----|----|----|----|----|---
       |                |
       useful protein   pointless proteins

The trick with the TACC being that perhaps TAAC, AACC, and TAGC all, rather than encoding a pointless protein, each encode a protein that will kill the cell. Any random damage to TACC will have a high chance of producing a sequence that will be deadly to the cell.

So each time you get whacked by a gamma ray, you spin a die and see what it hits. Maybe it hits a naturally occurring sequence like 'ACGATTACGATACG' and maybe you get cancer. Or maybe it hits one of those artificially inserted TACC segments that is designed to usually break in such a way that the cell is killed. Add more and more TACC segments and you become less and less efficient, but the odds get stacked more in your favor the more you add.

It would kind of be the "electric fence" approach to arresting unintended behavior (http://en.wikipedia.org/wiki/Electric_Fence).

I suspect actually pulling this off isn't something that is doable in the foreseeable future, since it would require being able to make strong statements of the nature "This segment of DNA should have no effect", and similar, but it makes sense in my head at least ;)

>If you have a species in a high-radiation environment, is it not possible that it evolves to have fewer reproductive mutations in general - and therefore lower chance of cancer?

IIRC there were studies of Chernobyl cows which showed that in such conditions genotype evolves such way that random mutations have less effects on phenotype, at cost of loss of specialisation.

I found that article, it's in Russian:

http://www.hij.ru/read/issues/2010/may/957/

I read it on elementy.ru:

http://elementy.ru/lib/431228

That only applies to base likelihood. Radiation adds a whole new level onto that by altering the structure of the DNA when it hits it. You could completely eliminate inheritable chances of cancer but still end up with cancer patients as a result of radiation poisoning,

Reply on news.ycombinator.com