Evolving Guide Trees in Progressive Multiple Sequence Alignment

We present a novel application of genetic algorithms to the problem of aligning multiple biological sequences through the optimization of guide trees. Individual guide trees are represented as coalescing binary trees which provide for efficient and meaningful crossover and mutation operations. We hypothesize that our technique avoids the limitations of other heuristic tree-building techniques, and is therefore capable of producing better trees, as measured by higher quality alignments. Further, our approach is more scalable than commonly used progressive alignment techniques when aligning large datasets.