Update Teaser - Gametes

Here anything Cell Lab related that doesn't fit into the other topics can be discussed.
User avatar
Petter
Site Admin
Posts: 668
Joined: Mon Dec 08, 2014 11:03 pm

Re: Update Teaser - Gametes

Post by Petter » Mon May 29, 2017 3:02 pm

@Bwisialo:
Your analysis is correct. From some tests it seems like evolution rather quickly gets rid of the gamete unfortunately. It would be good if this could be avoided somehow to allow us to evolve sexually reproducing organisms. One idea I thought of was to have something like a telomere length set by the gamete. After fusing this would be the telomere length in the resulting cell, it could be either an integer or inf. Every time the cell splits it would decrease the telomere length by 1 and if it reaches 0 it dies. The telomere length could not be allowed to mutate to inf (but to other numbers), since that value is a bit artificial. This way it would be impossible for evolution to get rid of sexual reproduction :)

@Megathosto14:
I'll make it a bit bigger.
Sakaaaaa
Posts: 197
Joined: Mon Jun 06, 2016 1:34 pm
Location: New Celland

Re: Update Teaser - Gametes

Post by Sakaaaaa » Tue May 30, 2017 4:22 am

So the combination is completely random? You're just swapping modes right? E.g. Org1 has a red M2 and Org2 has a green M2 and when they fuse M2 will be either green or red like that right? Also whats it going to be called? Gamecyte? Or just plain Gamete

Petter what about you hold a competition in the custom challenges section for a stemocyte level and the winner's challenge gets put into the game as the level to unlock the gamete? That would be cool. Gamete comes after stemocyte right?
"Only two things are infinite: the universe and photocytes on a sunlit substrate. I'm actually not sure about the universe!"
-Avert Eyntsein
User avatar
bwisialo
Posts: 2955
Joined: Sat Aug 15, 2015 10:22 pm

Re: Update Teaser - Gametes

Post by bwisialo » Tue May 30, 2017 8:28 am

It will be called a Gamete. The usable -cyte words, like gametocyte, are actually other things. Yes, that's how the genomes will combine.
amor fati
User avatar
bwisialo
Posts: 2955
Joined: Sat Aug 15, 2015 10:22 pm

Re: Update Teaser - Gametes

Post by bwisialo » Tue May 30, 2017 8:43 am

Would the telomere idea effect anything other than preventing the evolutionary elimination of gametes?

I had an idea about combining some suggestions, but the combining them might make them less appealing than they are separately. :P I can see that if gamete fusion brings about mutations as well as recombination, the mutation aspect can be redundant with radiation. Radiation mutations are memory-less and I've mentioned that it seems like evolving smart organism would be more efficient if there were mutations without wholesale memory loss -- mutations that, eg, changed a mode setting by a value of < or = x. If gamete fusion brought about this kind of mutation, it wouldn't be redundant and it would make the gamete feel like a significant reward for completing so many challenges. :)

Maybe I should just give up the ghost. :P
amor fati
User avatar
bwisialo
Posts: 2955
Joined: Sat Aug 15, 2015 10:22 pm

Re: Update Teaser - Gametes

Post by bwisialo » Tue May 30, 2017 9:13 am

Megathosto14 wrote:I really think that this will work just like virus, but in random way and all modes included.
Which is to say, it will work a lot differently than a virus. :P
amor fati
User avatar
Petter
Site Admin
Posts: 668
Joined: Mon Dec 08, 2014 11:03 pm

Re: Update Teaser - Gametes

Post by Petter » Tue May 30, 2017 10:32 am

The telomere would be just as I described it so its only effect would be to limit the number of mitotic divisions of a cell derived from a gamete fusion.

What do you mean by memoryless mutation? What is x?

I do think recombination errors would be a bit redundant. For sure substrate radiation is still required since many organisms will not be sexually reproducing.

Only advantage I see with recombination errors is that mutations then only happen on reproducing cells. If a swimmers flagellocyte gets a mutation then it either does nothing, harms it or is beneficial. It is typically much more likely to be harmful than beneficial. Having a high radiation will then be harmful and start to kill off too many cells, that is why we can not have a very high mutation rate. Beneficial mutations to the flagellocyte aren't carried on anyways so all mutations given to the flagellocyte are actually either useless or harmful for evolution. Limiting mutations to reproducing cells would therefore make evolution more efficient. Consider an organisms using M different modes in total, N modes(and cells) in its adult stage, and having 1 reproducing mode(and cell) in its adult stage. Lets say all mutations only happen in the adult stage. Lets say 0<p<1 is the probability a mutation to a certain mode is beneficial when it is expressed. Let's call the mutation rate per mode r. The probability rate of having a harmful mutation from radiation is then r*(N-1+M)*(1-p). The probability of having a beneficial mutation being carried on is r*M*p.

For only recombination errors we would get harmful mutation rate r*M*(1-p) and useful mutations rate r*M*p. If we fix the harmful mutations rates to some max possible value, then we see that only recombination errors would allow a useful mutation rate that is 1+(N-1)/M times larger than for radiation only. For a typical large organism I guess M=10 and N=6, this means using only recombination errors would allow an evolution rate that is 50% faster. Note I made a few assumptions here and might have made calculation mistakes, please correct if something is wrong. Anyways, to my current estimates the effect would not be huge, though it would for sure be useful. Note that increasing the plate size from 1mm to 1.22 mm has the same effect (measured in in-game time). It would add a bit of complexity though, should the mutation rate be set by plate or by genome? Should it be allowed to mutate?
User avatar
bwisialo
Posts: 2955
Joined: Sat Aug 15, 2015 10:22 pm

Re: Update Teaser - Gametes

Post by bwisialo » Tue May 30, 2017 7:33 pm

Petter wrote:What do you mean by memoryless mutation? What is x?
On another thread you wrote:
mutations are "memory-less". Meaning the mutated value is independent of the previous. So e.g. a is set to anything within it's allowed range.
I understood this to mean that, eg, if M1 Nutrient priority is 2.15, a mutation can result in an M1 Nutrient priority anywhere from 0.10 to 10.00.

Memory-based mutation would just mean that the range where the mutated value could fall would be dependent on the current value. Scrap "x," because memory-based mutation could use a range defined by ratios rather than a fixed value of +/-x. So, for example, for Nutrient priority, the range of a mutation could be from "current value / 1.21" to "current value x 1.21." That is, the mutated value would be any value between 1.78 and 2.60 but would not fall outside that range in a single mutation.

My thought was that, when gametes recombine modes, some modes would undergo a mutation of this sort, though I don't know how to specify the rate of mutation.

This would be different than radiation and would dramatically increase the ability to evolve smart organisms.

I have to look at the other questions a bit more.
amor fati
User avatar
Petter
Site Admin
Posts: 668
Joined: Mon Dec 08, 2014 11:03 pm

Re: Update Teaser - Gametes

Post by Petter » Tue May 30, 2017 8:25 pm

Ok, I see what you mean then. Yea some of the settings are memory-less, typically the signal substance dependent stuff. I also think it could be an improvement to change it so that the value has some memory of the previous value, though I'm not sure whether it would "dramatically increase the ability to evolve". It's hard to judge how to do this though. Should we just allow +-5% difference, or maybe +-50%? I think that will be very important and it will depend on the property. I think it might be difficult to optimize. There can also be cases where small mutations kill the organism but a big one would be good. Consider a swimmer with the tail lik 160 degrees the wrong way, it will survive but not be very efficient. Changing it's direction like +-30 degrees either makes it spin even more, or it makes the tail appear inside the other cell and it dies. Changing the direction by 160 degrees however is very beneficial so in that case one can not do a series of smaller mutations.

Another benefit of gamete reproductive errors: if the error rate is a property of the gamete that can mutate, then perhaps one would not have to tune the radiation level at all but evolution would set it to the value that is currently best? It's hard to think about these stuff but I imagine it could work.
User avatar
Nayus
Posts: 679
Joined: Thu Oct 06, 2016 2:03 am
Location: Argentina

Re: Update Teaser - Gametes

Post by Nayus » Tue May 30, 2017 9:54 pm

What about leting that to the mode? Have a unique Setting for gametes that can control the amount of changes or error on the copy of the childs. The minimum would be 0% and you could set up a maximum. We talk about probability of course, but maybe 50% (0.5) as top value.

That way you cover both benefits at the same time without ading the disadvanteges of a fixed behavior.

Because I'd argue against a fixed positive value. You sometimes want a stable behavior to test equilibrium. Currently you just turn off radiation. If the gametes don't have this option and still are forced to change overtime, that would be impossible to turn off.
User avatar
bwisialo
Posts: 2955
Joined: Sat Aug 15, 2015 10:22 pm

Re: Update Teaser - Gametes

Post by bwisialo » Tue May 30, 2017 11:08 pm

Yes, it's hard for me to think through these things. :lol: So, yes, it's hard for me to say how much more efficient evolution would be. Just to make sure there wasn't confusion on this part, I meant it would increase the ability to improve smart organisms via evolution, not that it would increase the ability to evolve smart from dumb. Actually, it would be more accurate to say signal-using organisms rather than "smart": standard crawlers (with Neurocyte and no Stereocyte) are some of the most difficult organisms to improve via evolution.

I'm thinking about cases where you are evolving a signal-using organism that is already set up in a desired or appropriately functioning way. The smart swimmer is not going away from food; the crawler's Myocyte is not turned sideways; etc. In general, I'd say these sorts of organisms are such that changing one setting by a large amount has, on average, a greater potential to be harmful than in other organism I've used evolution with. Large changes to multiple settings or multiple modes is a different matter. However, evolution is unlikely to yield changes like that unless it is via small changes that don't kill the organism, small changes that can accumulate over time.

One large change to a setting that determines the amount of substance produced or determines how the substance is used is also very unlikely to be beneficial. If you have an already functioning settings, you least of all want Stereocyte Output to switch from positive to negative; or the Output substance to change; or a Neurcoyte change that stops the oscillation, or Contraction to switch positive to negative. In sum, the multi-cellular, reproductive, and programming demands of these organisms seem to increase the rate at which single large mutations are harmful. Large change should still be producible, but you can get to them in small steps.

I think the range for Split mass and Nutrient priority could work in the proportional way I indicated. The range for Split ratio could be defined by a fixed value: eg, ratio to child 1 could be limited to a change +/-.05. Angles could also be defined a fixed value. Stereocyte Output could be +/- 10%: +/- 2.50 when Output is 25; +/-.25 when Output is 2.50. It would take a while for small value to change, but that ultimately seems for the best. Changes to "a" and "b" could also be a small percentage, perhaps less than 10%, but I'd need to look more closely at the numbers. Changing these in a Neurocyte easily breaks the oscillation.

A changing error rate makes my head hurt. :P
amor fati
Post Reply