Automata Chemistry
https://github.com/franticspider/stringmol.git
Hello. You have found brief instructions to install, compile and run stringmol version 0.2.2. Please read carefully, and mail any comments or questions to sjh436@gmail.com.
If you are new to stringmol, please also see the technical report "smtr0.2.pdf" included in the same directory as this document
Type the following at the command prompt:
sh install
Stringmol only builds in linux. Maybe one day we'll make it work on other platforms, but to be honest we are more interested in deploying it on the web and running experiments. email sjh436@gmail.com if this policy is causing you problems.
To compile stringmol on your system:
cd src
make all
should build an executable called stringmol in that directory. Do NOT run stringmol in there - it'll just make a mess. Instead, read the next section and do what it recommends.
It is NOT a good idea to run stringmol from the directory it was compiled into, since a lot of files will be generated and it'll be hard to work out what was generated and what is already there. The best thing to do is to create a folder for runs, and execute stringmol by entering the relative path to it.
Let's assume that we are starting from the directory containing this README. We recommend you try the following:
mkdir test1
cd test1
../src/stringmol
If all goes well, you'll get the following output:
USAGE for TestSM
The first argument after TestSM should be the trial type,
followed by the arguments needed to run it
TRIAL TYPES LIST (NUMBERS IN BRACKETS ARE EXPERIMENTAL!)
NAME TRIAL No: ARGUMENTS
1 on 1 0 .conf (.mtx)
ALife XII 1 .conf (.mtx)
Con Pop 2 .conf (.mtx)
Origlife (3)
Comass GA (4)
Joinsplists (5)
Energetic ALXII (6)
swdist (7)
Comass ALXII (8)
speigmonst (9)
Check setup 10 .conf (.mtx)
Finished!
This shows the current ways that stringmol can be run. More details below.
Firstly, it is important to know that not all of the trial types above actually work. If you know a little C/C++ programming, you can look through all the options if you want, or mail sjh436@gmail.com for more info. In this section, we are going to go through the "tested" options - those not in brackets in the list above.
The general method for running these trial is to type the following
PATH/stringmol TTYPE ARGS
where 'PATH' is the relative or absolute path to the strignmol-exe folder (e.g. '/home/simon/stringmol'), 'TTYPE' is the type of trial you want to run (e.g. '1'), and ARGS are the other arguments that are needed to run that particular trial. Arguments in brackets are optional.
For the currently working trials, the arguments are all the same. The first argument is the path to a config file that contains the details of the trial, and the second optional argument is to an 'mtx' file - the substitution matrices that are used in calculating alignments and creating mutations. A couple of examples of these files are contained in the 'config' subdirectory. We will one day make a guide to these files available. In the meantime, email sjh436@gmail.com with any questions.
This option is there purely to check if your config file has been read as you intended. It is a new feature in stringmol0.2.2.
To test this out, navigate to the folder containing this README, and type the following:
$ mkdir test10; cd test10 $ ../src/stringmol 10 ../config/test_replicase.conf
The output should look like:
Hello World! Let's run the stringmol tests!
Argc = 3
BEFORE loading the config, params are:
Non-stringPM variables:
NTRIALS not set - the default value would be used if needed
NSTEPS not set - the default value would be used if needed
CELLRAD 2500.000000 (vcellrad = 0.000000)
AGRAD 10.000000
ENERGY 0
NSTEPS -0.000012
BLOSUM 0 size table loaded
MUTATE indelrate = 0.000000; subrate = 0.000000
DECAY 0.000000
MAXLEN 2000, (maxl0 = 2001)
ESTEP 20
..c'est ca!
NTRIALS not specified;
Setting NTRIALS to 1
NTRIALS = 1
Setting NSTEPS to 1200000000
NSTEPS = 1200000000
CELLRAD = 2500.000000
AGRAD = 10.000000
ENERGY = 0.000000
NSTEPS = 1200000000.000000
ERROR 60 on loading table
No mutation rate found in config file
Using ALifeXII values instead
AFTER loading the config, params are:
Non-stringPM variables:
NTRIALS 1
NSTEPS 1200000000
CELLRAD 2500.000000 (vcellrad = 2500.000000)
AGRAD 10.000000
ENERGY 0
NSTEPS 1200000000.000000
BLOSUM 0 size table loaded
MUTATE indelrate = 0.000100; subrate = 0.000100
DECAY 0.000237
MAXLEN 2000, (maxl0 = 2001)
ESTEP 20
..c'est ca!
Finished!
If you open the file config/test_replicase.conf, you'll see that the values in the config file correspond to the entries following the text "AFTER loading the config..". This is handy for checking any config files that you build.
A new feature with version 0.2.2 is the ability to specify the substitution matrix at the command line. This was done to make stringmol available on youShare more easily - the .mtx file can be specified directly, rather than being fetched out of the .conf file. This may be easier for some manual configurations too. Note that since this modification was designed for youShare, runs of stringmol will not be interactive where the extra argument is used to specify the .mtx file - this is only important for TType 0, where the user presses the space bar to step through a reaction - if three arguments are present, the 1 on 1 simulation just runds until limits are reached (see below)
This option allows you to see how one molecule reacts with another. To test it, starting from the stringmol directory where this README is located, type the following:
$ mkdir test;cd test $ ../src/stringmol 0 ../config/test_1on1.conf ../config/ALXII.mtx > out.txt
This will create a file called 'out.txt' that will show each step of a reaction between the two molecules listed in the file test_1on1.conf, located in the config directory.
Should you want to step through the reaction, you'll need to enter the correct file path for the .mtx file in the .conf file that you use - see the line beginning with the keyword "SUBMAT" in 'test_1on1.conf'. Once this is done, the command is:
$ mkdir test;cd test $ ../src/stringmol 0 ../config/test_1on1.conf
you can then "step through" a reaction between molecules by repeatedly pressing the space bar.
We'd like to improve this facility. Please send feedback to sjh436@gmail.com
This is the grand-daddy stringmol program - running a "container" of molecular reactions together. We have included a config file with the same configuration as for the paper "ALxii_dif.pdf" that is included in the tarball.
$ mkdir test;cd test $ ../src/stringmol 1 ../../config/ALXII.conf ../../config/ALXII.mtx
This is an unpublished trial type, in which a set of containers of molecules are run. If the population of molecules in a container goes extinct, half the contents of another randomly-selected container are moved into the empty one.
You can test this trial type using the same configuration for TTYPE 1 using:
$ mkdir test;cd test $ ../src/stringmol 2 ../../config/ALXII.conf ../../config/ALXII.mtx
Currently only four containers are run, but we have done trials where there are 16 (C programmers can set this by changing the value of NCON on line 1565 of the file TestSM.cpp, and recompiling).
This trial runs fairly reliably, but there may need to be changes in the output files to make sense of what is going on. At the moment, the output to screen after the config files have loaded looks like:
0 0 1 2 3
Count: 150 150 150 150
Printing species list
1000 0 1 2 3
Count: 165 172 172 164
2000 0 1 2 3
Count: 198 222 214 181
3000 0 1 2 3
A list of species that the simulations have created is printed every 10000 time steps - the line "Printing species list" indicates this. Apart from this, there are alternating lines. The first line (e.g. "2000 0 1 2 3" indicates the timestep, and the number of each container. The second line (e.g. "Count: 198 222 214 181") indicates the total number of molecules in each container.
There are a number of issues that need to be addressed to improve stringmol. These features all currently work 99% of the time, but for longer simulations, the reacions explore edge-cases in the stringmol language that aren't fully explored/documented/understood. Consequences are:
- Repeatibility: Although we are recording the system state and the state of the RNG, restarts in deeper runs do not appear to be repeatable
- Logging: Species lists are fairly comprehensive, but if a molecule is created just before a run terminates, its origins may be lost
- Zero percent binds: in deeper runs, a situation emerges where a small percentage of reactions (typically about 0.1%) have zero probability of binding
SOFT_SEARCHpoorly definedHSearchappears not to be stochastic at present