Sun Aug 27 16:11:35 PDT 2000	Kevin Karplus
>T0121 MalK, Thermococcus litoralis

	Blast gets one good hit: 1b0uA
	double-blast gets one good hit: 1b0uA
	
	There are 3253 sequences in the t2k alignment, only
	one of which is in PDB: 1b0uA	3.31.1.11.1
	
	The 1b0uA FSSP file indicates some structural similarity fo
	1cr1A, 2reb, 1d2nA, 1skyB, ... but these seem to be only
	1-domain matches.

	The 1b0uA match is for the N-terminus, up to about residue 240.
	
C-terminus discussion:	
	At the C-terminus there seem to be blast matches to 1do2A,
	2ki5B, 1qhiB, 1ki6A, 1vtk, 1ki2B.  
	1dhy seems to span the gap.

	I created a T0121-cterm subdirectoy, and did a search using it.
	Indeed blast gets a number of weak hits.
	double-blast gets a couple of modest hits: 1eps and 1ffkQ.
	
	The t2k alignment for T0121-cterm gets 79 sequences, none in PDB.
	The target model finds two PDB sequences
		1b9n[AB]	21	(FSSP rep 1b9mA, not in SCOP)
	
	The best template models for T0121-cterm are
	Structure	TemplateReverse	TemplateEValue	SCOP
	1ef8A	-8.11	2.62244115102756		?
	1bncA_1	-7.87	3.33350443440372		2.79.2.1
	1psrA	-7.64	4.19513441802921		1.41.1
	1alvA	-7.32	5.77618916748044		1.41.1
	
	The best 2-track scores for T0121-cterm are
	% Sequence ID   Length      Simple     Reverse     E-value      SCOP
	1b9mA             265       -40.22      -23.95     3.5e-07	?
	1a02N             301       -33.60       -8.89     1.1e+00	2.1.1,2.2.5
	1theB             260       -23.21       -8.78     1.1e+00	4.3.1.1.11
	1theA             260       -22.43       -8.14     1.1e+00	4.3.1.1.11
	
	The 1b9mA/T0121-cterm-1b9mA-2track-local alignment looks quite
	good, with nice striping of conservation across the barrels.
	I think we can predict the C-terminus with this.
	
	We might want to splice together the two 2ry-structure
	predictions as well.

Mon Aug 28 12:08:30 PDT 2000	Kevin Karplus

The best hit with the target model is 1b0uA (of course), though there
are some fairly strong scores for 1f2u[BD] and 1f2tB.

The 1b0uA hit is a 2-way hit.
Other good template hits (actually 2-way) are 
	1cr1A
	2reb (2reb_1)
	1ng1
	1ckeA
	3adk
At least the first two of these have significant structural similarity
to 1b0uA.

The alignment 1b0uA/T0121-1b0uA-global looks pretty good.
We should probably cut it off at about residue 240 (P or YDKPANT), and
use the c-term prediction after that.

It might be a good idea to use Melissa's tools to get a consensus
alignment between the 1b0uA and T0121 based alignments.



Tue Sep  5 10:35:24 PDT 2000
remaking 2track

Wed Sep  6 13:31:56 PDT 2000	Kevin Karplus
The T0121-1b0uA-local and T0121-1b0uA-2-track-local alignments agree
(except for where the domain ends).

I'm building a t121-1-230 set of alignments, using "family" options to
try to get a smaller (though probably still huge) multiple alignment,
and to have something to paste together with the cterm, which starts
at 231.	HMM---hmmscore failed with a floating exception.



I've created what I think is the correct alignment for the first
domain in 
    		1b0uA/1b0uA-T0121-karplus.a2m
And I like 	1b9mA/T0121-cterm-1b9mA-2track-local.pw.a2m.gz
for the second domain.

The secondary structure predictions should be spliced together from
first 230 of T0121.t2k.2d
and	t121-cterm/T00121-cterm.t2k.2d
I have put this spliced file in 
	T0121.t2k-spliced.2d