package kappa-agents

From KaSim 3 to KaSim 4

The Kappa syntax has completely changed

We are sincerely sorry for the inconvenience. We strongly believe that the new syntax will be less error prone.

For legacy kappa code written in ye olde Kappa, the KaTools can be called from the command line with the option -syntax 3. The output log file "inputs.ka" will be a copy of your model, but updated to KaSim4 syntax.

Comments follow the C standard

• // start a line comment

• /* and */ flank a block comment

• # no longer starts a line comment; it is the new wildcard for site state

Sentences split over several lines no longer require a backslash \

You can go to a new line anytime anywhere, don't put \. For example:

%mod: [T] > 500 do (
    $ADD 50 A() ;
    $DEL 20 B() ;
    $SNAPSHOT "snap.ka" ;
    )

Site specification uses new symbols

The general picture is now the following:

• linking states are specified between square brackets []

• internal states between curly brackets {}

• internal states cannot be integers anymore

• free sites have an explicit syntax: [.]

• the syntax for whatever is # instead of ?

Here is a correspondence table:

Agent order in rules must be maintained

There is no longest prefix convention anymore. The i-th agent on the left hand side must correspond to the i-th agent on the right hand side of the rule. There must be exactly as many agents on both sides. Agents degraded by the rule are replaced by a dot . on the right. Conversely, agents synthesised are replaced by a dot on the left. For example:

// Agent A spawns a copy of agent B
A(),. -> A(),B() @ 'r'

// When dimerized, agent B degrades agent A
A(x[1]),B(x[1]) -> .,B(x[.]) @ 'r'

If a site's internal state is modified, it must appear on both sides in arrow notation. I.e. Agents must specify the same sites and the corresponding states on both sides. For example:

// Change from state u to state p
A(x{u}) -> A(x{p}) @ 'r'
// Force internal state to change from whatever to p
A(x{#}) -> A(x{p}) @ 'r'

You must put an explicit # in A(x{#}) -> A(x{p}) to say that you wrote a force internal state change on purpose (that's why the "whatever" # has been introduced for).

Certain uses of comma are now optional

Commas in internal state lists, or site lists, are optional. E.g.

%agent: A(x{u p b q} y z)

Commas remain mandatory in lists of agents. E.g.

A(x[.] y[.]), B(x[.] y[.]) -> A(x[1] y[2]), B(x[1] y[2])

Tokens in a list are separated by , instead of +

 | 1 U -> | 1 Kr, 1 Ba @ 'r'

Colon : no longer separates tokens and their quantity

This change also standardizes the syntax for initial conditions between agents and tokens. The syntax to introduce tokens is: %init: [alg_expr] [token_name]

%init: 5 A(x[.])
%init: 9 ATP

Unary rates in ambiguous cases are in {} instead of ()

In situations where a rule's left-hand side can be embedded both into a connected graph and a disconnected graph, two rates are required. Usage is:

[LHS] <-> [RHS] @ [binary rate] {unary rate}, [reverse rate]

Kappa speaks about intervention instead of perturbation

every modification (intervention action) ends by a semicolon

Interventions used to take a semi-colon separated list of modification as body. Now, every modification ends by a semi-colon. Said diferrently even the last (or only) modification of a intervention ends by ;.

%mod: |A(a)| < 100 do $SNAPSHOT "snap_at_".[T].".ka";

%mod: |A(a)| < 100 do
    $SNAPSHOT ("snap_at_".[T].".ka") ;
    $PRINT (|A()|." present at time ".[T]) > "myfile.txt" ;
    $ADD 10 A();

Repetitive intervention syntax has changed

Usage is now if precondition do modification(s) and repeat as long as condition.

%mod: [precondition] do [modifications] repeat [condition]

Beware: the second condition is the negation of the old until clause. I.e. it is a repeat as long as rather than the old repeat until this is true.

$FLUX has become $DIN

DIN stands for Dynamic Influence Network which is a more accurate name for what this modification generates

Syntax of $PRINTF has changed

The new syntax allows one to redirect output to a file.

$PRINTF "what you print" > "Where_you_print.txt"

The > can be thought as the shell redirect operator.

Syntax for "observables" is no longer a special case

The number of occurrences of the pattern pat is now written |pat| and it can be placed in any algebraic expression.

Thus, a variable is defined as:

%var: '[name]' |[pattern]|

For example

%var: 'o' |A(x[.]),B(x[.])| // Correct

%var: 'o' A(x[.]),B(x[.])   // Syntax error, lack of pipes
%var: 'o' A(x!1),B(x!1)     // Syntax error, lack of pipes and legacy bond marks

This changes allows more concise statements, for example

%var: 'o' 1 + |A(x!1),B(x!1)| / 2 // Oneliner in KaSim4 syntax

%var: 'tmp_o' A(x!1),B(x!1)       // Twoliner in KaSim3 syntax
%var: 'o' 1 + 'tmp_o' / 2

As a (more or less direct) consequence, you can write

%init: "any algebraic expression" "mixture"

%mod: $TRACK "pattern" [true];

Options -t and -e are gone, and -p has changed!

You should specify the limit of simulation by -l and the plot period (how often you want a data line) by the new -p. These options are by default in simulated time unit but you can switch to event-based using -u event.

The motivation for this change is interactivity. If you don't specify any limit, the simulation will run forever (use Ctrl-c to stop it).

Data files are printed in CSV (comma separated value) format

Previously, the default was to produce an data.out file with space separated values. The default is now to produce a csv file data.csv.

In addition to comma separated files, KaSim can also generate tab separated files (.tsv), and even a vector-based graphical representation of the time traces, indexed by time (.svg). This is done by specifying the file extension of the output file with the option -o [filename].

The motivation for these changes is to deal with weird values (e.g. infinity, not a number) in a way understood by (at least) GnuPlot, Python, Matlab, etc.

$STOP without argument does not dump snapshot

Use $STOP "[filename]" to get a final-state snapshot with the right filename.

Ctrl-c does not kill the simulator

Ctrl-c now pauses the simulation and launches a toplevel environment in which you can do modifications interactively, and then (optionally) continue the simulation.

You can use -mode batch to get a Ctrl-c that does kill the simulator (and more generally to get a KaSim that never asks questions and assumes default answers) or -mode interactive to pause the simulator just after the initialization phase has completed (i.e. before the first event happens).

Concretely the old behavior of Ctrl-c and then y(es I want a snapshot) is recovered by Ctrl-c and the modification $STOP "dump.ka"