Introduction To SIS-I
SIS simulates a primitive synthetic immune system. The SIS is "born" as if in an embryo and it is forced to make a self-nonself discrimination before being allowed to respond to external nonself antigens. To keep the interface simple in Normal Mode (not Expert Mode) we have "hard-wired" a test of your rules that includes making a self-nonself discrimination. Then if that test is OK there is an automatic test of an immune response starting some time well after birth and lasting for 7 days. If at the end of these preliminary trials there are 400 - 500+ effector B cells, then the simulation can be called a success. However, if you are interested in following other rules for making the self-nonself discrimination, use the expert mode. In the expert mode you are presented with a newborn immune system and by adding antigen immediately you simulate the response to a self-antigen. If you delay the addition of antigen, then you simulate the response to a nonself antigen.
The elements of this immune system are simply T and B cells. Antigen specificity is reduced to a single parameter, and for the time being it is fixed as a frequency of antigen reactive cells (3 per 1000 for B cells and 30 per 1000 for T cells), and the justification for these round figures is given below. The simulation covers a sample of one million cellular slots that are filled with either T or B cells that are either antigen-specific or not. The antigen is considered to be present uniformly throughout the million cells unit and any reaction that depends on antigen is switched ON or OFF by simply setting the antigen added or removed switch. There is no opportunity to encode the behavior of non-specific elements such as antigen-presenting cells and their products. Although Protecton Theory requires antigens to be recognized via multiple paratopes to induce the aggregation of soluble Ig, SIS treats these as a single specificity. A future model of SIS will allow multiple epitopes and multiple paratopes per antigen in order to incorporate more of the subtlety of Protecton behavior, along with antigens that grow and are eliminated by the immune response of SIS.
A set of default rules has been provided to show how T cells and B cells might behave in the presence and absence of antigen. The antigen that is added and subtracted by the operator, and only those T and B cells that can react with that antigen are tracked. As some places it is suggested that you could choose to add either intracellular or extracellular forms of antigen, but for the time being only the intracellular form functions in SIS-I. To simulate extracellular antigen an arbitrary decision can be made to render the TK system unable to respond to the extracellular antigens. In the simulator, this can be perfomed by removing rules that have TK cells along with antigen as conditions; thus eliminating the TK responce to the antigen. In summary, two forms of antigen are designated intracellular and extracellular, and these do or do not respectively activate the TK system.
The T and B cells are divided into three different states designated i, a, and e. Cells of any type or state are allowed 8 age steps that can be tweaked to an almost infinite number by tricking the rules (see later how this is done). The T cells are divided into two classes, one that is regulatory called H for helper, and another that is destructive called K for killer. For rather arcane reasons another state called x was added to the set of possibilities for B cells to allow a simulation of memory B cells.
Rules describe the conversion of cells from one to another. Basically you have a cell with inputs that converts to another cell state with outputs. A list of these rules is called a rule set. The SIS program applies the best matching rule in the rule set to each cell in the system. Hence, SIS is indifferent as to the order of the rules in the ruleset. In the case where there are two equally matching rules, probability is used to randomly utilize each rule half the time.
Hard wired into the program are some important limitations. Primarily, the effector functions of eTH are limited to those cells of any type that are within a chunk of 100 total cells. This effectively simulates a cell-cell interaction boundary for eTH. However, the eTH is able to secrete cytokines called IL-1 and IL-2 that act over a chunk of 10,000 total cells and any cell within that domain can respond to the secreted chemokines.
The whole system functions as a set of nested cellular automata with one million total cellular elements. This was a limit decided by the results of our studies on the behavior of a Protecton, which in essence simply states that there is a minimum number of cells that can be taken from an immune system which can mimic the behavior of the total. The concept of a Protecton arises from the elephant-tadpole paradox, which states that an immune system is made of an iterative unit that is present in varying copy number in differently sized species and differently sized individuals of different ages.
On the left of the "=>" separator is a description of the starting state of a cell along with any conditions needed to effect a change in that cell. Each additional condition is separated by the "<" punctuation mark.
An implicit feature in every rule is that each cycle of the automata is the equivalent of a cell division, and at every cycle the cellular objects increase one unit of age.
The consequence of the stated input condition is given following the "=>" separator and if there are multiple consequences, these are separated by the ">" punctuation.
Not always is there only one possible output, and to accommodate this it is possible to distribute the outputs into three probability categories. If there is a distribution of possible outcomes this is indicated by preceding the description with P(0.x) such that the sum of the probabilities is unity. At the time of execution of the rule in the program one of the possible outcomes is chosen so that at different times different outcomes are produced with a probability defined by the rule.
The best way to understand the system is to simply run the default rules offered and then look over the entire rule set and begin making changes. Designing a new set of rules from scratch is difficult unless the vocabulary and structure of the rules system is well understood.
Here is a picture of the SIS Rule Authoring Tool, with each button and input field labeled for easy reference.
To write a rule using the interface, press on the 'Cell' button (1). The following window will appear:
Select a starting cell State, Type, and Age. Press the 'Accept' button when done, or the 'Discard' button to cancel. Next press the 'Condition(s)' button in the SIS Rules Authoring Tool to select input conditions associated with this rule. The following is an example of the Input Conditions window with Antigen selected.
Many input conditions may be selected, or none. Not all of these input conditions have the same scope. Antigen if present in the simulation is present throughout the system. Effector T-Helper and effector T-Killer cells are limited to cells within a contact group. Similarly, IL-1 and IL-2 are limited to cells within a hormonal group. Notice that 'Input' box (A) will contain the 'convert from' component of the rule. This component can be cleared using the 'Reset' button (B).
Having completed the 'Convert From' component, move onto the 'Convert To' component of the SIS Rules Authoring Tool. Begin by pressing the Cell button (3). Just as before, a cell definition window will appear. Select the resulting cell State, Type and Age. Then press 'Accept' when done.
Now press the 'Results' button to obtain the following window:
The output of the cellular transition may be nothing, the production of antibody, a cellular division resulting in two identical cells, secreted IL-1 and/or IL-2, or any combination thereof. After making the output selection, press the 'Accept' button.
Insert a probability value of this output into the probability box (6), and then press the probability button (5) to update the 'Convert To' component of the rule displayed in (C). If the probability was 1 then you may skip the rest of this paragraph. If the probability is less than 1.0 then select a cell (3), results (4), probability value (6), and press the probability button (5), in that order. Do this until the sum of the probabilities is 1. If you wish to restart the 'Convert To' section, you may press the 'Reset' button (D).
Once the two parts of the rule has been constructed. Add the new rule to the rule set by pressing the 'Add Rule' button (7). The new rule will be added to the end of the rule set. Do recall that the order of the rules in the rule set does not matter, as the best maching rule is used for each starting conditions.
To delete a rule, fill in the rule number in box (9), and then press the 'Delete Rule' button (8). If the rules are miss-numbered because they were manually modified, pressing 'Delete Rule' with the default value of "N-N" causes NO deletions, but does cause the rules to be re-numbered. Now, if you want to delete a rule, say 15, then insert 15-15 and press 'Delete Rule'. A block of rules may be deleted, by providing the range. For example a value of "15-20" will delete rules 15 to 20, inclusive.
Unless you have your own account, Load Rules (11) will not be useful. Simply stated, it re-loads the last rule set that was submitted. As there are many users using "guest" your saved rule may be lost. If you use this page often and you want to keep your rules, just copy them off the web page, save them in a text file, and paste them back in when you are ready to go again. If you want to play seriously, please email firstname.lastname@example.org and ask for your own private sandbox to play in.
Add and remove rules until you are satisfied, or just leave the default rule set to see them in action. These initial rules are to get you on the right track, but you should test you imagination. Press the 'Submit Rules' button (10) to send the rule set to the SIS program.
Finally, it is acknowledged that the rules are primitive and deal
with a minimal set of features of the immune response. However, to have
a system such as SIS that starts with an embryonic immune system and produces
protective levels of antibody in a sufficiently short time to be protective
against infection is a good starting point.
The special set of rules, just the first three lines, is sufficient to cause autoimmunity. Usually it does not occur, but because the system is based on random arrangements, sometimes a new simulation will develop autoimmunity in less than 50 days. Using the Expert mode of the program skips the autoimmunity test of the rule set, and allows the user to specify the timing of the presence or absence of antigen from initial conditions.
Some basic graphing of AG, iTH, eTH, iB, and eB is performed to give the user a quick overview of the simulation. In a future version of SIS, we will allow the user to select the variables they desire to graph. For the time being the raw data is also available. Thus, if more graphing of variables or clarity is desired, the user can feed the raw data into their favorite graphing program.