Comment by bloak | Hacker News Reader

Comment by bloak | Hacker News Reader

Parent Full thread

Even if it were the case that each replication made the chromosomes smaller (personally I would guess that there is a non-zero probability of them not getting smaller and even a non-zero probability of them getting larger thanks to a mutation) that would still only imply a hard limit on the number of replications. You would also need to find a hard limit for how long a cell can survive without replicating. Good luck with that.

I content that there are no limits: however old you are, you have a non-zero chance of surviving another picosecond. The product of a finite number of non-zero real numbers is non-zero, therefore...

Some people may wish to retort that at some point these numbers become so small that they are no different from zero in practice. Those people are probably not mathematicians. They're not wrong, either, of course, but they're not making a valid argument for there being a limit.

There's no limit for the number of successive heads you can get when tossing a coin, either, though you won't in practice see more than a hundred if it's a reasonably fair coin. If there's no limit in that simplified, mathematical situation, does it make sense to hypothesise that "there is a limit" in the case of highly complex biochemical processes?

This is all very philosophical. It doesn't help us to estimate the chances of a given person who is alive and 115 years old today (there seem to be about five of them) one day breaking Jeanne Calment's record of 122. Are any bookmakers taking bets?

>Even if it were the case that each replication made the chromosomes smaller (personally I would guess that there is a non-zero probability of them not getting smaller and even a non-zero probability of them getting larger thanks to a mutation) that would still only imply a hard limit on the number of replications.

The fact that chromosomes get shorter during DNA replication is a scientific fact. Chromosomes have padding at the end called telomeres that provide a buffer from the 'useful' DNA. There is an enzyme that can lengthen telomeres, and its of great interest to scientists interested in extending human life.

There is almost no chance that mutations:

a) regularly occur in such a way that makes the chromosomal DNA longer, especially in a way that offsets the base pairs lost during every transcription.

and b) occur in the telomere area of the genetic code, after all the useful bits of DNA.

If mutations did occur at a rate that would counteract the shortening of the telomeres, the rest of your genome would be mutating so quickly you would almost certainly die rather quickly.

There are a whole host of other reasons why what you wrote is nonsense, and I don't have the time or the space to address them all. There is literally no fact to what you have written. It is clear you don't have even a rudimentary understand the biology of genetics. An understanding of math doesn't preclude the need for understanding the actual mechanism of how things work.

The following articles are a useful primer to understanding the flaws in your reasoning:

Okazaki Fragments: https://en.wikipedia.org/wiki/Okazaki_fragments

Telomeres: https://en.wikipedia.org/wiki/Telomere

Telomerase: https://en.wikipedia.org/wiki/Telomerase

DNA Polymerase: https://en.wikipedia.org/wiki/DNA_polymerase

DNA Replication: https://en.wikipedia.org/wiki/DNA_replication

That's all very interesting, but...

> almost no chance

That's my point, really. It might not be a terribly interesting point, but you're conceding it, not refuting it.

Does genetics actually provide a useful estimate of the probability of someone reaching the age of 123 years? I mean: something a bookmaker could use? Is there any evidence that telomeres are practically relevant to the longevity of humans in particular, as opposed to organisms in general, some of which live very much longer than humans?

I humbly suggest that you turn down the pomposity a tiny bit. Read what you wrote there:

> There are a whole host of other reasons why what you wrote is nonsense, and I don't have the time or the space to address them all.

What kind of impression do you think you're making?

I majored in Biology, with a focus in genetics and computer science because I wanted to go into bioinformatics. I could literally write 30 pages about why what you wrote is incorrect, and it would take several orders of magnitude more time than it took you to write out your hypothesis.

I provided resources for you to educate yourself. If you believe that it isn't a good use of your time, that's exactly how I feel about addressing the points you made beyond broadly saying 'this won't work, here are resources that address this on a level which you can understand.'

In this case 'almost no chance' is somewhat analogous to shuffling a deck of cards and finding them in order by suit and value, and then shuffling them again and finding them in reverse order by suit and value, and repeating that feat 10 times over.

The length of a human's telomeres when they are born is about 11k base pairs. Chromosome 21 is the shortest chromosome, and has 46.7 million base pairs. That means with random chance a mutation is 4200 times more likely to occur in the coding region of the chromosome than in the telomere.

Due to the way DNA Polymerase works, you will lose 20 base pairs of DNA on every replication. Ignoring everything about rate of mutations and the likelihood of insertion mutation, this means your chance of lengthening your chromosome through mutations is (1/4200)^20, or 1 in 2.6x10^72.

The odds of shuffling a deck of cards and having it come out in suit and value order is somewhere around 1x10^68.

That's using a best-case scenario as an example. Chromosome 1 has more than 5 times as many base pairs as chromosome 21, and you'd literally need to have this happen on every single chromosome every single time you had cell division.

A quick follow-up to this point. The exome, or coding region of the DNA is approximately 1% of the genome, so arguably the chances of a mutation randomly occurring in a coding region are 1/100 as probable as I suggested here.

There are a few caveats:

1) The non-coding region appears to be less useless than previously assumed. There are still 'highly preserved' areas in non-coding regions. If a section of the genome is highly preserved, it means that a mutation in that region probably results in death/non-viability of the organism.

2) We know the rate of mutation of the genome. If random mutations were really adding enough base pairs in the telomere region to lengthen it, the genome would be growing at an incredible rate.

There are a lot of reasons why it's also implausible, but they have to do with the amount of energy in a bond, etc. and other biochemistry stuff that I'm not qualified to comment on.

I mean, there is a non-zero chance of basically everything interesting, but that is meaningless. You're right that those who don't acknowledge this are not mathematicians. On the other hand, any mathematician with a passing interest in calculus realizes that the limit of the product of an infinite number of non-zero numbers less than one is exactly zero. To use your own phrasing, those that cling to these essentially meaningless likelihoods are probably not familiar with calculus.

That being said, the shortening I am referring to is the simple mechanical fact that we know with absolute certainty that your chromosomes shrink by 6 base pairs at each division. This is a simple biological fact.

There are a finite number of protein molecules on a human, and a finite number of humans. Given a available days, ut is impossible to assign an accurate arbitrarily small nonzero probability of incremental survival.

"Not being mathematicians" is exact what you want when making predictions about the real world.

Zeno's Paradox was solved like 300 years ago. Its just calculus.

Reply on news.ycombinator.com