Another pointless evolution program

[Blog of Reason]

Discussion thread for the blog entry "Another pointless evolution program" by rabbitpirate.

Permalink: http://blog.leagueofreason.org.uk/science/another-pointless-evolution-program/

 

[e2iPi]

I just let it run for a few hours, nice work. Only suggestion I have would be to some type of physical barrier between populations in order to show how speciation works on isolated populations.

-1

 

[darthrender2010]

lol, the population evolved into boobcreatures. Joking aside, very cool

 

[MRaverz]

lol, the population evolved into boobcreatures. Joking aside, very cool

Same happened here, I guess it's to do with their ability to get 'food' more effectively.

I really like these programs, although I found the information regarding what each button did and how it was relevant to be lacking or too wordy.

 

[Aught3]

This is awesome.

You should check out Darwin pond for some seriously cool evolution simulation.

 

[Rockbottom87]

haha, my little buggers died :twisted:

 

[Heavenly Spoon]

Always fun to play with these

Turns out size does matter; 1st generation had 12 lines and 1 stomach:
http://img687.imageshack.us/img687/1082/screenshot20100118at074.png
Strangely, none of these ever evolved colour, though they did evolve quite a bunch of red food digestions, no idea what that's all about...

For this one, every gene had +1:
http://img8.imageshack.us/img8/1082/screenshot20100118at074.png
Not as immense as the other one, but at least they have some colour...

My regular tests went way too slow, and some of them even went extinct, natural selection is not kind towards these litter critters

 

[Rockbottom87]

http://img687.imageshack.us/img687/256/evolprog.jpg

At first I thought the food had grown to immense sizes, and my buggers died again somewhere during the night.
But then I remembered there were no blue foodlings.

 

[rabbitpirate]

Turns out size does matter

Hmmm, maybe I should put some kind of limit on how big they can get so that things don't get completely crazy.

 

[rabbitpirate]

I really like these programs, although I found the information regarding what each button did and how it was relevant to be lacking or too wordy.

Could you be more specific and I will see what I can do to make it a bit clearer.

 

[MRaverz]

I really like these programs, although I found the information regarding what each button did and how it was relevant to be lacking or too wordy.

Could you be more specific and I will see what I can do to make it a bit clearer.

My main problem was that I didn't really have a full understanding of which gene did and whether putting a greater value to it would be beneficial (and in what way).

Red multiplier and red suppression for example, I still have no idea what it is that they do exactly.

 

[rabbitpirate]

My main problem was that I didn't really have a full understanding of which gene did and whether putting a greater value to it would be beneficial (and in what way).

Red multiplier and red suppression for example, I still have no idea what it is that they do exactly.

Yeah I get that. I wanted to put better descriptions but was a bit limited for space. I think I will add another page to the About section that better explains what is going on. Thanks for the feed back.

 

[monitoradiation]

if evolution is really like this program, why wouldnt creationists believe in evolution? c'mon its so fun god MUST'VE played with it <3

 

[Catch22]

Very interesting program, RP. It reminds me alot of the swimbots program, but I think in many ways it's a better model of DNA mutation's role in evolution. In your simulation you have inconsequential genes (color / non-coding), slightly advantageous genes (having more sides), very advantageous genes (digest red food), and essential genes (having stomach).

In your model you have 4 DNA bases, just like real DNA. Instead of codons (blocks of DNA code) being organised into groups of 3 bases, they are organized into pairs. So, instead of 64 possible code combinations, there are just 16. They are as follows:

AA - Non Coding
AG - Side
AC - Longevity
AT - Non Coding
GA - Red Multiplier
GG - Blue Multiplier
GC - Green Multiplier
GT - Stomach
CA - Red Suppressor
CG - Blue Suppressor
CC - Green Suppressor
CT - Non Coding
TA - Red Digestion
TG - Non Coding
TC - Non Coding
TT - Non Coding

So you can see, there are 10 codes for various genes and 6 non-coding combinations. Unlike real DNA, there are no Start or Stop codons, so it seems the entire code would be used in this simulation, but I have some questions.

Questions:
Here is the DNA code for the starting creatures broken down into codon pairs:

AG,TG,TA,CA,GC,CG,CG,TG,AT,TG,AA,TT

You notice the first 2 bases [AG] would code for a side. The second [TG] is non-coding, but the 3rd [TA] would code for digestion of red food. If this is the case, then why do none of the first generation shapes possess this ability? I encounter similar confusion when I compare my analysis of the DNA to the number of coding genes in the description. So what am I not getting here? [Edit: I think I figured this one out... read at bottom]

My second question is about longevity [AC]. Is it cumulative (2 [AC] genes lives 2x longer) or does only one copy have any benefit?

Observations/Predictions:
I have three general predictions about what I should expect to see evolve in this simulation under the rules I've just described. All have to do with odd numbered additions or deletions resulting in the DNA code shifting.

The first thing, is that I should expect the DNA code to steadily grow larger. This serves the creature by increasing the possibility that if a code shift occurs and a beneficial gene code is distorted, a replacement gene code might arise from another part of the genome. For example, the Stomach gene code is essential for reproduction. If we have the code:

AG,TA,GT,GC,CA,GC,CG,CG,CC,AT,TG,TT,AA

...you can see the last 4 pairs in this sequence are non-coders. But, lets say the next generation suffers a deletion of the T in the [GT] gene, responsible for the stomach. The genome then looks like this:

AG,TA,GG,CC,AG,CC,GC,GC,CA,TT,GT,TA,A

Here a new [GT] stomach gene has been created from the non-coding genes. Since there aren't really any detrimental gene codes, gene additions are always beneficial to the creature, even if it's non-coding.

The second thing, I believe because these odd numbered additions/deletions affect all the code after them, that the most essential genes will usually appear at the beginning of the code as a safety measure against code shifting. The stomach gene [GT] is the most important , followed by the red food gene TA, followed then by the side gene [AG] and the longevity gene [AC]. I hypothesize that shape creatures will evolve to mirror this gene order to maximize their chances of survival.

Third, I predict:

1. Early evolved shape creatures will need a high thymine [T] count
2. Later creatures will find more adenine [A] to be beneficial.

This is because Thymine is needed to code for stomach and red food digestion, and multiple copies of this base will help ensure survival if an essential code is changed, like in the example given in my first prediction. However, Thymine has a value limit. Once There are a sufficient number of copies of the gene to ensure against a fatal mutation, then additional Thymine bases lose there benefit, since having additional stomach/red food genes has no addition benefit for the creature. Assuming the longevity gene's [AC] benefits are accumulative like the side gene's [AG], then having lots of adenine bases in a sufficiently large genome would probably be very beneficial. The more adenine, the more sides you have and the longer you live, the more prolific you are.

I haven't had time to test some of these ideas. Let me know what you think of them, or let me know if I've misunderstood something about the simulation.

Suggestions:

• Pause Button
• Put a scrollbar on the description sidebar
• # Side genes = # of sides (it's a bit confusing)
• Commas to break up the codes in the DNA, or anything that helps to understand how the DNA code relates to the genes
• Ability to Copy DNA from and Paste DNA code into the simulation
• There should be some gene codes with deleterious/negative effects
• If you build a new version, it would be nice to see what effect a 3-base code system would have (ex: [ATG],[CCA])
• It would also be interesting to see what effect start/stop codons would have on this system

[EDIT: I think I understand the code better now, and it seems it IS infact a 3-base code system, but with the third base basically ignored, AFAIK. This might invalidate some of my predictions, but I'm not sure how exactly it will effect them]

 

[AndromedasWake]

RP I've just had time to check this out, and although I haven't run it very long, I can already see it's great! Have you thought about talking to Spork about a LoR skin for it? We could put it in the site's resources.

Also, can you compile it as an application to run natively on Windows/OS X? Macs handle flash quite badly, so it knackers the CPU to run something like this and heats up the computer quite badly :roll: :roll:

Great work man. These projects are excellent teaching aides to show people evolution in action.

 

[rabbitpirate]

Very interesting program, RP.

Why thank you. Ok let's have a look at your questions. Firstly, and looking at the program this is entirely my fault, I think I have given the wrong idea about how the DNA works in this program. I think I tried to be too clever with it and then failed to explain exactly what I was trying to do.

The genome in the program is divided into sections of 3 letters rather than 2 letters. This was intended to simulate the way that real nucleotides work in DNA. So yes you are right that AG codes for sides, but more accurately it is AGT and, as in real DNA, also AGC, AGA and AGG. All of these code for sides.

What this means is that mutations in the program can be neutral as a mutation that changes AGT into AGC will not affect the number of sides the shape has. I'm sure with that bit of information the rest will make more sense as you clearly have a firm grasp on the subject. Other than that small misunderstanding I think your predictions and pretty much spot on.

Just to cover your other question quickly. Yes the longevity gene is cumulative but does not have the effect of doubling the life of the shape. Basically the average shape lives for 25 shape years, less if it can't find food. If it has one longevity gene this increases to 26 shape years, 2 genes 27 shape years etc etc.

While I am at it let me cover the issue that MRaverz raised re what exactly the colour suppression/multipler genes do. The way I ended up having to do colour is a result of the way that computers create RGB colours. I wanted the shapes to basically start out with no colour and become more colourful over time. The problem is that the RGB value for white is 255,255,255 - as high as the values go.

As such rather than simply having genes that increased the amount of, lets say, Red the shape had I was forced to do it the otherway as 255 is as high as the red value goes. Instead the genes subtract from this value. So if you have a section of the genome that looks like this:

CAGCAC

Then you have two genes for suppressing the red value, making it 235. This results in a colour for the shape that is more a mix of the blue and green than it is red and so it comes out this colour.[

The colour multipler basically increases the effectiveness of the suppression gene. So a genome of:

CAGCAC

Results in a red value of 235. However if you throw in a multipler gene for a genome of:

CAGCACGAT

This will multiple the effect of those red suppressor genes so that you get a red value of...er...215. Basically one multipler gene multiples the effect of the suppressor genes by 2, 2 gene multiples by 3 etc etc.

Anyway I hope that makes more sense now. Oh and thanks for the suggestions Catch22, I will have to give them some thought.

 

[rabbitpirate]

RP I've just had time to check this out, and although I haven't run it very long, I can already see it's great! Have you thought about talking to Spork about a LoR skin for it? We could put it in the site's resources.

Also, can you compile it as an application to run natively on Windows/OS X? Macs handle flash quite badly, so it knackers the CPU to run something like this and heats up the computer quite badly :roll: :roll:

Great work man. These projects are excellent teaching aides to show people evolution in action.

Thanks for that TK. I would definitely be up for sticking the LoR brand on it if you guys are happy to do so. I have to say that this program is very much a proof of concept thing rather than a glossy product, but hey if people think it would be useful then I am more than happy for them to use it as a teaching aid. Maybe I can combine both this and the last one into one program and stick the LoR logo on it. Hmmmm

As for porting it to run directly on windows etc I have no idea how to even start doing that, but sure yeah I'd have no problem with that if someone showed me how.

I'm trying to think of a way to make it more of a game rather than just a sim. Not sure how it would work yet but it might be cool if people could design things and compete them against one another...but as I said I have no earthly idea how to even start doing that so don't expect anything like that in the near future.

 

[Aught3]

Anyway to male them breed? It might help to sex the game up a little...

 

[ImpudentInfidel]

Fun program. Ran a few at the same time to see the different routes. One thing I saw was that batches would stagnate for hundreds of generations with little variation, although I note these were the ones with almost no non-coders which I suppose makes sense. (Edit: A later run proved that idea wrong.)

A way to speed up the simulation like in your first would be appreciated, so that these periods don't take hours to get out of. It's not like anyone's going to sit and watch that whole time, but still...

I think as a game it would be fun, but it would also undermine the concept. Put in much more control than it already has and it becomes an Intelligent Design sim instead of an evolution sim. Combining it with the bug program would be pretty cool though.