Author Topic: Epigenetics and Viruses  (Read 4023 times)

Offline Endy

  • Bot Overlord
  • ****
  • Posts: 852
    • View Profile
Epigenetics and Viruses
« on: February 14, 2006, 03:43:38 AM »
After talking with Elite about it, I sat down and I finally managed to get bots that  safely have relativly stable random gene transfer.

Managed to find and overcome most of the problems, so far. One of the main ones is that veggies can pick up animal dna and become evil animal plant hybrids. Managed to prevent veggies from using the animal dna with epigenetics. The animals all store a value into 971. Then something like:

20 *.vel sub 2 div .up *971 777 div mult store

is used to prevent all these actions from occuring. A bit modified from what Elite mentioned, but the idea of preventing genes from activating in another bot really helped.

The second major was the reverse of this, animals acquiring plant dna. This led to the bizzare rapid reproduction I'd been seeing in the sims. I switched the plants dna to use basically the same repro gene the animals used and it was no longer a problem. I attempted epigen for plants, but many were dying off too fast to transfer the information(non-reproing plants are not fun).

Yet to be solved:

The worst problem left is caused by how relativly unintelligent the bots' genes are. Normally bots have one maybe two genes per action at most. Now imagine there are some 100 genes many exactly identical... The population would rise normally for awhile then begin to decline, apparently the additional gene costs tacked onto the gene transfer costs, were simply too much and eventually wiped them out. Fortunatly a simple "This location already has a value" dna strand should solve this one. :)

The second major problem, one I never thought of is that smaller genes replicate faster. This led the most useless of genes to gain an ever increasing percentage of the genome. To combat this I think a cyclic virus creator gene is in order, this way no one gene would gain an unfair advantage.

Huge genomes:

I'm planning to add in a "genome too big" deletion gene, keeping the genome down to a random 50 genes.

Alright, all I've got for tonight. Kind of makes me wonder though if real dna tries to keep its secrets to itself similarly. Makes sense that gene transfer might not be such a good thing between competing species.

I want to keep working on the dna some, I'll hopefully have a fully working bot in the next few days.
« Last Edit: February 14, 2006, 03:51:27 AM by Endy »

Offline Ulciscor

  • Bot Destroyer
  • ***
  • Posts: 401
    • View Profile
Epigenetics and Viruses
« Reply #1 on: February 14, 2006, 11:02:41 AM »
Kind of makes me wonder how real dna can distinguish between different species. Of course I'm literally stupid about this subject.

However... reading a bit of the last post... "smaller genes replicate faster"... seems v v interesting to me, as I remember hearing about lengths of dna who's sole purpose was to get copied, something like genetic parasites. Also, shorter dna lengths are subject to lower probability of errors.
:D Ulciscor :D

I used to be indecisive, but now I'm not so sure.

Offline Elite

  • Bot Overlord
  • ****
  • Posts: 532
    • View Profile
Epigenetics and Viruses
« Reply #2 on: February 14, 2006, 11:20:59 AM »
Well done Endy, sounds good.

The main way real-life orgainisms acchieve horizontal gene transfer is through sexual reproduction. You are right - organisms do not share with other species. Bacteria and other simple celled organisms do I think via plasmids.
I think we should ask Shvarz ...

Random Ideas:

Maybe you could use ties to 'steady' the bot and hold a conspec in place throughout the gene transfer?

Is there any way you could get bots to rate their own fitness and display this. Ie. "I've got superior genes to you, come and get them". out1?

What about randomly .vshooting genes, but when your genome gets too big (>50?) you just delete a few genes?

How can read DNA distinguish between species? It doesn't  :P
Gene parasites ... interesting.

Offline Endy

  • Bot Overlord
  • ****
  • Posts: 852
    • View Profile
Epigenetics and Viruses
« Reply #3 on: February 19, 2006, 01:31:50 AM »
Quote
What about randomly .vshooting genes, but when your genome gets too big (>50?) you just delete a few genes?

Yup, this worked the best. Kept them around 30, and they survived for a good long while, without the odd rise and drop effect. Haven't had the time to do as much with them as I'd like. Work's been getting in the way. -_-  I'll start them up again tonight and see what happens. They eventually died out last time, still not sure why...

For transfer, the bot's normal conspec avoidance genes are getting in the way somewhat. Maybe make them stop rotating away from a conspec when they have a virus ready.

Quote
lengths of dna who's sole purpose was to get copied

Could experiment with adding a dna'less gene to see what happens. Presumably this would have the best chance of being transfered.
« Last Edit: February 19, 2006, 01:41:03 AM by Endy »