BIOMURFFS


The program is designed to demonstrate how a succession of small changes, when
stacked on top of one another, can lead to radical differences in the system
affected. This is used by Richard Dawkins in 'The Blind Watchmaker' where he
uses it to demonstrate the principle of cumulative small change as a mechanism
for Darwinian evolution. There are one or two differences between Dawkin's
program and this, but these are relatively minor ones connected with the
operation and not the overall effect.


SYSTEM REQUIREMENTS

The program is written in TURBO BASIC V1.0. It needs an IBM PC or compatible
with CGA or EGA graphics. The memory requirements I don't exactly know about
since the smallest system I've run it on is 512K. The program loads the DOS
utility GRAPHICS.COM to allow screen printing, but this is called with no
parameters, so it is set up for IBM GRAPHICS / EPSON compatible printers.
The CGA high resolution graphics mode is used for plotting the biomorphs, so
you don't need a colour monitor. Anyone who wants to write in genes for colour
on an EGA is welcome to do so.


USER INTERFACE

Run the program from DOS by entering BIOMRF from the DOS prompt and pressing
the RETURN key. The screen will clear and the title page printed. At the foot
of this you are asked:

Do you want to define the initial gene values yourself ?

answering Y will allow user definition of the intial values - see later.
Answering N will cause the screen to clear again and the following to be shown

The default initial values are:
6,2,8,1,2,135,90,10

Do you want these or a random selection ?

See the section on genes for what this means. If you want the default values,
just press the RETURN key, if not, the starting values will be generated at
random, and you will be shown what they are. You are then told that to leave
the program, just press the RETURN key in response to the question:

Please enter the number of the biomorph you want ?


DISPLAYING THE BIOMORPHS

The messages here are pretty self-explanatory, with the 'Do you want' type as
Y/N replies and the others requiring numeric answers. However:
1. Notice that the morphs are plotted 1 2 3 4 5 6 7 8
9 10 11 12 13 14 15 16
but there may be fewer than 16 there. You can only select morphs which are
actually shown. If you try to select number 8 when there are only 6 plotted
you'll get nonsense. They are plotted staggered for clarity.

2. When you enter the magnification, values MUST be positive. Magnifications
less than 1 reduce the size of the morph. As the morphs have a limited range
of sizes and the screen has a limited size, you're best using magnifications
of between 0.5 and 3. The gene values are plotted to the side of the morph if
you ask to see it alone.

3. If you press RETURN on it's own to the question:

'Please enter the number of the biomorph you want ?'

you will leave the program.


BIOMORPH GENETICS

OK, this is where we see how it's all done. In natural systems, genes are just
data storage media which, via the mechanisms of transcription affect (directly
or otherwise) visible properties (or phenotypes) of the organisms, WITHOUT any
obvious connection between the gene and the phenotype (i.e unless you know the
system you don't know what one set of data in a particular gene will do just
by looking at it). This is what happens here, but the system for turning gene
data into phenotypes is simple enough to do on paper.
Mutation is alteration of genetic data resulting in an altered phenotype. With
natural mutation, very often the most obvious phenotype is death, but enough
non-fatal mutations go on (especially with the added source of variety from
sexual reproduction) to allow development. Biomorphs do not die, nor do they
reproduce sexually, so in order to get evolution fast enough to see, they have
a huge mutation rate - every one is mutated. However, each mutation is in only
one gene, and it is only a +/- 1 change in the value of that gene. Since all
these small changes are cumulative, a lot of development can be seen.




I've just mentioned what biomorph mutation is physically, so to put it into
perspective, it must also be said here that biomorph genes are all members of
the set of natural numbers. These are thus data (in the same way as the 4- way
'numeric' code of DNA is data) which are read by the relevant section of the
program as an instruction as to how to do it's job. The main difference (other
than one of scale) between biomorphs and living organisms is that in biomorphs
the phenotypes are produced directly from the code.

Each biomorph has 8 genes of this type. As has been said, one is mutated at
each reproduction, but which one is chosen at random, and the mutation, +/- 1
is also decided at random. The non-random influence here which can lead to the
development of particular phenotypes is human selection. The genes are:

1: Number of offspring.
2: Number of iterations used when plotting (see PLOTTING BIOMORPHS)
3: GENE # 1 for branch length
4: GENE # 2 for branch length
5: GENE # 3 for branch length - if odd, branch length is 3 * 4, if even branch
length is 3 + 4.
6: Angle of 1st branch from precursor (in degrees).
7: Angle of 2nd branch from 1st (2 branches come from each branch point).
8: Length of initial branch (trunk).



PLOTTING BIOMORPHS

Biomorphs have a plane of symmetry running down the middle. This both looks
nicer and is easier to do as it needs less data. They are actually 'trees'
with a trunk and two branches coming from each branch point. They are plotted
as a series of vectors, so the data needed to plot each one are co-ordinates
of the start position, the length & the angle of the vector from the external
vertical/horizontal x/y co-ordinate system of the screen. As there are two new
branches coming from each old one, at every iteration after the first two an
additional 2^(iteration) vectors are generated (remember that only half of the
biomorph is actually worked out, as the other half is plotted by reflecting
each vector on this side across the mirror plane), so it is easy to calculate
where each new vector came from. (The file BIOPLOT.BAS contains the subroutine
which does this). Once the vector is calculated, it can be plotted along with
it's mirror image, and the subroutine for this is in the file VTRPLOT.BAS.






The vector is calculated by finding it's start co-ordinates, which are the end
of the parent vector. The length of the new vector is found by combining the
values of genes 3 & 4 in the way indicated by gene 5. This is then stored and
it's angle is found from the angle of the parent branch and gene 6 (for the
first branch from a given branch point, or from the angle of the first branch
and gene 7 for the second.

This process is repeated for the number of iterations given by gene 2, and the
biomorph thus drawn. The same is done for each biomorph in the set. When you
ask for a single biomorph to be plotted, this again happens, the only
difference being that the length of the vector is multiplied by the value of
the magnification given.

Biomorphs can also be plotted onto the printer at any time by pressing

PrtSc.


I am currently writing version 2.0 of this program, which will include two
competing species and a definable environment. If you want a copy of this,
once it is finished, including source code, or if you just want a chat, my
email address is:

[email protected]



D.J. Murphy, 12th May 1988