Before you go investing your future earnings :-),
I suggest you look into the VAST literature on this
subject. Due in part to the Human Genome Project, there
is enormous interest in automatic base calling, and I suspect
that every pattern-recognition algorithm has been tried on
the problem. In particular, you might do a literature search
for Clark Tibbets (sp?) or Lloyd Smith, both of whom have worked on
neural net approaches.
It just ain't a trivial problem.
Good luck & enjoy
Keith Robison
Harvard University
Department of Cellular and Developmental Biology
Department of Genetics / HHMI
robison at mito.harvard.edu