#Please insert up references in the next lines (line starts with keyword UP)
UP	arb.hlp
UP	glossary.hlp

#SUB	subtopic.hlp


TITLE		Optimize database compression

OCCURRENCE	ARB_NT

DESCRIPTION	Sequence data normally need's a lot of memory. To be able to 
		handle thousands of sequences we implemented an online
		compression. All data is compressed most of the time and only
		uncompressed on demand. As a user you only find smaller database
		files, that's all.
		Without understanding the data, the program can compress data only
		by a limited factor. With the help of a tree aligned sequences
		can be compressed much better by storing only the differences
		to a consensus sequence.
		Once a sequence is compressed using a tree, it will keep
		the good compression method until it is changed. Then only the
		older method is used.
		As long as you change only a few (up to 100) sequences, the
		database won't grow very much.

		To compress the entire database, the program needs a tree,
		which should cover most of the sequences. The larger and better
		the tree, the better the compression.

EXAMPLE		10000 aligned 16s sequences need 50 mega-bytes of memory.
		Without your help ARB will reduce them to 10 mega-bytes,
		and given a tree not more than 2 mega-bytes will be needed.


NOTES		Any major database update, especially inserting or deleting
		gaps in an alignment, should be followed by a new optimization
		step.

BUGS		No bugs known