Thu Jul 13 11:49:06 PDT 2000	Kevin Karplus

>T0113  Short chain 3-hydroxyacyl-coa dehydrogenase, Rattus norvegicus
	tetramer, no disulphides, known homologs
	molecular replacement on 2hsd

	Blast finds excellent hits for 2hsd[ABCD] and 1hdc[ABCD]
	(several other almost as strong hits)
	
	double-blast finds excellent hits for 2hsd[ABCD] and 1hdc[ABCD]
	(several other almost as strong hits, including
	1ah[hi][AB] 1fmc[AB] 1aei[AB]
	
	The top hits all have FSSP rep 1bdb (with identity just 30%),
	though 1cydA is the rep for many of the almost as strong hits
	(with identity just 31%).
	
	We should probably use 2hsd, since the crystallographers did
	molecular replacement using it.
	
	The t2k alignment has 1357 sequences in it, including 10
	distinct PDB sequences, so we probably want to use a
	subalignment.  Phytree is probably going to be unable to make
	a tree for such a large set of sequences though.

	Using the FULL t2k multiple alignment, the best alignments are
2hsdA/T0113-2hsdA-global	2hsdA             253      -179.31     -187.47     1.2e-81
2hsdA/T0113-2hsdA-vit		2hsdA             253      -180.51     -175.60     1.6e-76
2hsdA/T0113-2hsdA-local		2hsdA             253      -183.54     -172.60     3.3e-75
2hsdA/2hsdA-T0113-global	T0113             261      -142.83     -152.87     1.2e-66
2hsdA/2hsdA-T0113-vit		T0113             261      -142.75     -136.86     1.1e-59
2hsdA/2hsdA-T0113-local		T0113             261      -147.70     -136.59     1.4e-59
	
	The 2hsdA/T0113-2hsdA-global alignment looks excellent, with
	three short insertions and one deletion, all at the extreme
	ends of loops.  The first 5 residues are not matched, then the
	insertions are IKTYHE, NEPDQG, P (in qvpfPsrlg), and the
	deletion is 
		VKLLSDtsSYVT
		QMVIEN..PFLN
	
	The 2hsdA/T0113-2hsdA-vit alignment is the same, except for
	not aligning the first 6 residues or the last 6.

	The 2hsdA/T0113-2hsdA-local alignment is the same, except for
	not aligning the first 6 residues or the last 5.

	The 2hsdA/2hsdA-T0113-global alignment has slightly different
	insertions:	
		AIKTYH	(one earlier)
		PDQGGQ  (two later)
		pfpSrlg	(one later)
	and the deletion (sd) is two residues earlier.
	Also the first and last 5 residues are not aligned.
	I like the gap placement of the T0113 model better.
	Again, the vit and local alignments agree except for slight
	differences in how much is unaligned at the ends.
	
	Overall, I like 2hsdA/T0113-2hsdA-global best.

Wed Jul 26 16:36:21 PDT 2000	Kevin Karplus
	Actually phytree did make the tree, but the drawing packages
	can't draw it.
	
	There are multiple sets of PDB structures in the tree, all
	apparently in SCOP family 3.2.1.2.*
	
	We want to decide which of the subfamilies is the closest to
	the target, so we can get the best template.
	Based on the BLAST scores, we get a fairly clear ordering:
	3.2.1.2.12	this is the one 2hsdA is in.
	3.2.1.2.11
	3.2.1.2.18
	3.2.1.2.19
	3.2.1.2.13
	3.2.1.2.20
	3.2.1.2.6
	3.2.1.2.14
	3.2.1.2.10

	
27 July 2000
	For completeness, ran the normal SAM scoring on the target,
	as expected the top hits are all 3.2.1.2 family.
	The best 2-way hits are 1hdcA and 2hsdA, followed by 2ae2A,
	1ybvA, 1bdb, 1cydA, 1a4uA, 1b16A, 1fds, 1oaa, 1eny, 1d7oA,
	1enp, 1zid, 1dhr, 1udc, 1xel, 2udpA.
	
	Check to see whether then 1hdcA and 2hsdA alignments are different
	from each other, and choose the more appropriate template.


31 July 2000 kevin karplus
	
	Christian and I created T0113-2hsdA-global-edited.a2m,
	combining our preferences from the T0113-2hsdA and 2hsdA-T0113
	alignments.