From: Jorja Henikoff <jorja@muller.fhcrc.org>
To: compfly@currant.lbl.gov
Subject: BLOCKS - Documentation file
Date: Sat, 26 Jun 1999 11:08:18 -0700 (PDT)

6/26/99	Fly Annotation Experiment - BLOCKS

Method:

1. Searched 2.9mb DNA sequence vs Blocks+ (15June99, http://blocks.fhcrc.org)
and vs blocks extracted from Smart 3.0 multiple alignments 
(http://coot-embl-heidelberg.de/SMART/) using BLIMPS, which translates
the DNA sequence in all 6 frames for comparison with blocks.

2. Post-processed BLIMPS search results with the "blkprob" program to compute
expected values, using combined (for multiple blocks from the same family) 
cutoff evalue = 10. Sorted results by location on each strand.

3. Removed the following types of results:
	a. Unsupported alignments to single blocks known to be compositionally
	   biased (not supported by other blocks from the family).
	b. Unsupported alignments to single blocks with evalue > 1.

4. Manually investigated regions where blocks from different families hit
in the same region.

5. GFF file fields: There is one record for each block, so if multiple
blocks were hit for a family there will be multiple records.
	1 Sequence name = Adh_and_cact.1
	2 Method = BLOCKS
	3 Type = similarity
	4 Start aa of alignment with block
	5 End aa of alignment with block
	6 Evalue for single block (this is not the combined evalue
	  for multiple block hits)
	7 Strand
	8 Frame
	9 Block name, description, "number"
	  More documentation can be found at:
	  http://blocks.fhcrc.org/blocks-bin/getblock.sh?<block name>
	  The "number" groups multiple block hits.
	  
-------------------------------------------------------------------------
Jorja Henikoff 				           jorja@fhcrc.org
Fred Hutchinson Cancer Research Center  	   FAX: 206-667-5889
1100 Fairview AV N, A1-162, PO Box 19024  	   Seattle, WA 98109-1024