Agents Discussion

That sounds good. I'll edit in a table soon covering all the agents covered so far plus those, and that should be sufficient for first release.

where is the table it said soon over a week ago

Right here.

These are the finished agents so far.

Name	Target	Effect	Magnitude
Toxin	AI cells	Deals damage to cell membrane	= 35 - (35-15) * 1.3^(-n)
Paralytic	Flagella, cilia, and lamellipodes	Decreases efficiency	= 80 - (80-50) * 1.3^(-n)
Organelle Depressant	Mitochondria OR chloroplasts OR thermoplasts	Decreases reaction rate	= 75 - (75-30) * 1.3^(-n)
Organelle Stimulant	Mitochondria OR chloroplasts OR thermoplasts	Increases reaction rate	= 75 - (75-30) * 1.3^(-n)
Signal Agent	AI cells	Attracts or repels cells	= 35 - (35-5) * 1.3^(-n)

These (or this) are the agents yet to be specified in their implementation.

I'm not a programmer but how about periodic function, every 2 seconds or something like that, counts random number from 1 to 100 with condition AI cell is in signal agent region, lets say lvl 1 signal agent has 10% chance to attract or repel AI cells - every 2 seconds random number is chosen if its 1-10 the agent takes effect and AI cell is given order to move closer or further from the player?

Should we also add a neutralisation agent for each (an agent which prevents one of the enemy's from working)? I think it was discussed before, so the verdict may have been that these wouldn't be implemented.

By the way, thanks for the table. I'll need it for the animation. :) 

What about an agent that heals cells?

I think same. We will need a agent like that.

I agree with above, and maybe it could be called a regenerative, since it repairs the cell? (Wow, regenerative is a word)

Yeah that could work, I feel we need a counter to the poison and stuff, in order to balance it.

This is going mmorpg, i can't recall anything from biochemistry that is magically healing cells. And why would you want to heal AI cells anyway, if you mean it for the player the membrane should be repairing consuming ATP as it should be one of the basic functions of the cell as is moving.
Actually for repairing the lipid layers in the cell is responsible the smooth endoplasmic reticulum, by lipid layers i mean cell membrane and all the membrane restricted organelles - almost all of them except ribosomes and centioles.

Hmm I suppose. I was thinking for healing other cells of your species. But we might not even need that agent really. I dunno, see what people think.

I see, well in the end someone with authority over the project should decide, as healing agents maybe good from gameplay perspective after all we see something like that in every game but i think instead of healing them the player should be encouraged to just multiply more often. It's closer to the unicellular organisms logic and is scientifically more accurate, even our bodies prefer to just send signal for self destruction to damaged cell instead of repair it, cells that are infected with viruses or mutate into tumor cells are destroyed and broken to food for the other.

The thrive team has taken liberties before, with thermoplasts.

I agree with Tritium, I don't think healing agents are a good idea. Thermoplasts may be a fabrication, but they are not that far fetched. Buffs and healing powers, on the other hand, begin to wade into MMO waters.

Tritium wrote:

I'm not a programmer but how about periodic function, every 2 seconds or something like that, counts random number from 1 to 100 with condition AI cell is in signal agent region, lets say lvl 1 signal agent has 10% chance to attract or repel AI cells - every 2 seconds random number is chosen if its 1-10 the agent takes effect and AI cell is given order to move closer or further from the player?

By the way, Tritium, thats a good idea. That means Signal Agents can now be moved to the finished agents list.

Tritium wrote:

This is going mmorpg

This agent list went MMORPG a long time ago. So, in an effort to turn this thread back to science, here is a nowhere-near-complete list of some interesting real-life agent-likes off the top of my head:

• The immune complement; which is the most intricate example I know of a cell-bursting chemical cascade
  
• Mitochondrial toxins like tetracycline; with easy-to-understand in-game effects
  
• Ribosomal deactivation; particularly nasty
  
• A whole host of cytoskeletal agents; all of which stop the cytoskeleton from doing its thing and could thus disrupt cell-cell adhesion, motility, secretion, and the endomembrane system
  

No need to scrap the current list -- maybe some research will turn up some scientific basis for the agents we already have, which would make me one happy camper. I'd just like some agents to have more interesting effects that can teach some science.

Tritium wrote:

I'm not a programmer but how about periodic function, every 2 seconds or something like that, counts random number from 1 to 100 with condition AI cell is in signal agent region, lets say lvl 1 signal agent has 10% chance to attract or repel AI cells - every 2 seconds random number is chosen if its 1-10 the agent takes effect and AI cell is given order to move closer or further from the player?

I, a programmer, endorse your idea. Admittedly, I'd do it slightly differently in terms of how the AI cell responds, but that's OT.

NickTheNick wrote:

Thermoplasts may be a fabrication, but they are not that far fetched

Thermodynamics disagrees, but that's OT, and okay fine they're pretty cool. But still. Nyuh.

Oh oops I necro'd. Ah well, had to be said.

Don't worry about going OT or necroposting, what you suggest is very productive. 

If you have ideas for more realistic and engaging agents and agent effects, by all means post them here in the agent/target/effect/magnitude format. What you linked to so far looked good to me. '

Also, in your reference to how you would handle what Tritium suggested with the signal agents, I don't think you should worry about your idea being OT. Signal agents are quite important to discuss, especially considering that they are a component to creating a multicellular colony.

Ah k, thanks

Oh yes, about signal agents -- I read around and it seems like my implementation isn't very feasible, so I'd stick with something like Tritium's unless something better comes up.

Okay, so based off my agents above (well the new ones), but in detail:

-------
Cytoskeletal-destabilizing kinase: A protein that phosphorylates cytoskeletin (I made it up, shush), causing the cytoskeleton to dissolve. It renders victims almost immobile, unable to phagocytose or use contractile vacuoles, and almost completely incapable of secretion via secretory vesicles.
-------
Ribosome-inactivating protein: Eats ribosomes. Prevents protein synthesis. Causes slow deterioration as proteins aren't replaced. If the dose is too low to destroy all ribosomes, it just requires an energy-intensive period of recuperation after the effects wear off. Otherwise, fatal. Possibly too powerful to include in-game, but we could balance it as being hard to produce (so you'll usually administer too low a dose).
-------
So, we have:
CDA/cytoskeleton/immobility, slows secretions, phagocytosis and contractile vacuoles/Highly concentration-dependent, decays quickly
RIP/ribosomes/No protein synthesis, slow deterioration but certain death if dose high enough/has a toxic threshold dependent on victim cell size

And some science for the already-chosen agents:

A good match for mitochondrial depressants would be something like tetracycline -- in higher than medically-safe concentrations, it inhibits mitochondrial activity somehow (I'm fuzzy on details, lent my book on the stuff to someone who still hasn't returned it).

As for membrane damaging, it isn't as simple as just popping a balloon -- phospholipids in solution will quickly organize into sheets and vesicles, and will seal any gaps as fast as you make them.What you have to do is inflate the cell -- then it's like an over-inflated balloon, and it'll split right open. Oddly enough, you inflate it by poking holes in it, which is what the complement system and membrane-damaging agent, already do

More scientific backing for what you've done already will come as I think of it~

Some very interesting additions. I like the idea of each toxin having both a distinct way of harming other organisms (rather than reducing HP), and where possible a realistic basis.

In terms of some toxins being too powerful, this might be balanced by their potential effect on your own cell, unless we require that cells be resistant to their own toxins, though equally we can use the difficulty in acquiring/producing the toxin as balance. We need to consider how useful slow-acting poisons will be in a gameplay sense, as in most cases you'll use your toxin to weaken another cell and then either immediately run away or destroy it.

I suspect you mean tetracycline as ribosome inhibitor, as its function is to prevent amino acids binding to ribosomes during protein synthesis. A good mitochondria inhibitor would be cyanide, which blocks part of the electron transport chain in respiration.

I think it would make the most sense for things that would otherwise be susceptible to certain agents to only be resistant through evolution. However, evolving resistance should either be difficult, or be part of an arms race. Then, the easiest way to resist some agent toxic to chloroplasts would be to not have them in the first place, but if you absolutely needed them, you'd need to be able to evolve resistance fast enough.

You make a good point about slow-acting agents though -- if all they'll do is round out our agent selection, we can leave them for later.

And yes, I was misinformed -- my cytology book has been returned to me since and it agrees with you.

Not relying on whatever organelle is susceptible to a toxin would make for a very interesting game play dynamic, although some organelles are far from optional. Chloroplasts, flagella/cilia, your own toxin production, and maybe others would be viable things to go without, and you could endure for a very limited time without mitochondria or protein synthesis.

Perhaps slow-acting toxins are more interesting from a defensive view, as the player may need to endure these in order to attack another cell. Multiple weaker toxins might also interact to be deadly, if you don't have access to a single more powerful option.

Toxin resistance/potency are a great example of an evolutionary arms race, and also a good candidate for horizontal gene transfer (i.e.: plasmids), though as noted elsewhere that's more accurate in prokaryotes. I've seen various articles that suggest that both toxicity and resistance have intrinsic metabolic costs, even when they're not actively being used, which could prevent a cell from being resistant to all known toxins.

I had an idea a while back about nonspecific (multicellular stage) toxins and their implementation, which was that their efficacy in an organism depended on matching hex values, which takes care of the issue of being immune to all toxins.
Essentially, each toxin (here, agent) has a two-digit efficacy hex code (possibly linked to the agent's target, or just randomly generated) and each organism has a two digit resistance hex code. The lower the absolute difference between the two, the higher the resistance to the toxin/agent. This can be likened to an immune detection scheme and the toxin's shape being changed in order to go unrecognized by the cell. Not sure if this would be needed with the highly specific agents in unicellular stage, but it's food for thought if we want to get into how resistance works.

Also, we can incorporate cyanide as well to shut down mitochondria.

in terms of implementation, I'd lean more towards taking the bitwise xor of a resistance fingerprint and an toxin fingerprint. The more they match, the better the resistance. However, since resistance to a toxin doesn't necessarily lead to higher susceptibility to another, we can also generate a second bitmask for every agent to represent its mechanism. Then, we take toxin XOR resistance AND mask, and the higher this is (which is due to a combination of lower evolved resistance and wide-ranging effect (a mask with many ones)), the stronger the toxin. The mask never changes, and the toxin and resistance fingerprint are subject to auto-evo.

Extra note on my reasoning for a mask: In both your model and mine sans mask, evolving one bit of resistance to one toxin either raises or reduces your resistance to every other toxin. Every toxin is strongly coupled, when they really shouldn't be. The system naturally has one stable state -- for example, for two toxins, it has them at opposite ends of the spectrum and the resistance in the middle. We probably want to not have one stable state at all, and I think random masks are enough. All we'd need to do is some experimentation to figure out a happy average number of ones in the mask to make sure there's enough intertoxin interaction but not too much.

This sounds good, except I'm stumbling over your meaning here:

moopli wrote:

Then, we take toxin XOR resistance AND mask, and the higher this is (which is due to a combination of lower evolved resistance and wide-ranging effect (a mask with many ones)), the stronger the toxin.

Are you saying that you apply the AND operation on the already computed XOR for resistance/toxin fingerprints and the mechanism bitmask?
If I understand you correctly, the system works like this:
the cell has one universal resistance bitmask
each agent has a toxin bitmask

The match between those two determines how effective (efficacy) the agent is.

In order to preserve dynamic evolution, each agent also has a non-changing mechanism bitmask.

Final efficacy of any agent is determined by how many bits in the efficacy mask match the mechanism mask.

Is that right? That sounds good to me. It also creates an interesting and complex trade-off problem if we keep the same system controlling interactions with both harmful and helpful agents.

You have hit the nail on the head.

Especially with your final comment -- this is exactly what I want to see happen. Evolving specificity of pheromones, venoms hijacking signalling pathways, the works.