To build WAM matrices:

A multifasta file containing a consensus for the considered signal is required.

consensus = context before the signal + signal + context after the signal

Example for a virtual start file, with ATG and a short context (3+2):
>Start01
ATTATGCT
>Start02
TTTATGCA
>Start3
CACATGTT
...

Some parameters are also required (explained by a WAMbuild -h)
1 The order of the markovian models
2 The total length of the consensus
3 The position of the first nt of the consensus
4 name of the multifasta file
5 base name for the output

warning: 
Depending on the purchased order, some additionnal context could be
required in the multifasta file before the first nt of the consensus
(min length of this additional context = order of the markovian model)
Some possible commands for the example:

WAMbuild 0 8 1 start.true.fasta start.wam.true
WAMbuild 1 7 2 start.true.fasta start.wam.true 
(with an order 1, the consensus can't begin before the second nt to allow 
WAMbuilder to build the first model)

One model (binary file) is build per position of the consensus.
For the WAM sensors, two sets of model are required, one for the true
positive, and one for the false positive (at each signal occurence ATG,
GT or AG the log likelihood is computed).
A text output is produced with words occurence counts per position.
(false positive files can be produced by a simple extraction of ATG, Gt
and AG with some context).

