Parallel Application Systems in Genetic Information Processing
Masato Ishikawa, Tomoyuki Toya And Yasushi Totoki
Institute For New Generation Computer Technology
4-28-21f, Mita 1-chome, Minato-ku, Tokyo 108, Japan ishikawa@icot.or.jp
Abstract:
We have developed parallel application systems for genetic
sequence analysis. The systems are three parallel iterative aligners
and an alignment workbench for the aligners. The parallel iterative
aligners use a best-first search, a hill-climbing search, and a
genetic algorithm. Each iterative aligner can produce higher-quality
solutions than the conventional tree-based aligner. The genetic
algorithm shows better performance than the best-first and
hill-climbing searches. The alignment workbench, which features the
parallel iterative aligners, realizes alignment which is not only fast
and high-quality-it is also constraint-based. When a user has some
biological knowledge which indicates that some characters might be
aligned in a column, a constraint can be defined for those characters.
The constraint set is considered simultaneously in each iteration
cycle of parallel alignment. Then appropriate multiple alignment is
generated by the aligner and displayed on the workbench's full-color
display. The alignment workbench also contains the following
characteristic sequence analysis modules: a phylogenetic tree drawer,
a motif-database matcher, and a stem region specifier.
All of this proceedings (Compressed PostScript file) :
Parallel Application Systems in Genetic Information Processing (302KB)