Order to sort the input sequences for cluster_fast. Valid values are other (the default, meaning process sequences in the order they appear in the file), length (sort by decreasing length) or size (sort by decreasing size annotation).