8 June 2000 Kevin Karplus T0097 C-terminal domain of ERp29, rat weak BLAST hits for 1baz[ABCD] 1bdv[ABCD] strong double-blast hit for 1aj8[AB] weak double-blast hit for 1c1g[ABCD] 13 sequences in t2k alignment (none of them in PDB) All are in same family according to phytree. No hits with target model. No hits with template model. 9 June 2000 Kevin Karplus May have to try building HMM with more remote sequences, since this model didn't find anything. The remote model finds some weak hits, all to the same FSSP representative: chain E-value FSSP rep 1fkw 29. 1a4mA 1a4l[ABCD] 32. 1a4mA 1a4m[ABCD] 32. 1a4mA 1add 32. 1a4mA 2ada 33. 1a4mA 1uio 33. 1a4mA 1uip 34. 1a4mA 1fkx 36. 1a4mA The 1c1g[ABCD] sequences are just coiled coils, so all they are predicting is that T97 has a lot of half-exposed helices. Since we had already predicted that it was an all-helical protein (with some breaks in the helices), this is not very informative. The alignments 1aj8A/T0097-1aj8A-local.pw and 1aj8A/T0097-1aj8A-vit.pw are identical, but the global alignment 1aj8A/T0097-1aj8A-global.pw is aligned to the beginning of 1aj8A, which makes no sense, since it puts several beta strands where helices are expected. The local alignments predict helices, with the boundaries not too far from what the 2ry structure prediction has. The packing of the helices seems a little too loose though, without the rest of 1aj8A to hold them apart. It may be that t97 is a 4-helix bundle. The alignment 1a4mA/T0097-1a4mA-local.pw gets only part of the target aligned, and the viterbi alignment is even shorter. The helix predictions match fairly well, though we predict longer helices than in this alignment. Again the arrangement seems a bit loose without the rest of 1a4mA. Mon Jun 19 11:42:48 PDT 2000 Kevin Karplus Looked at CAFASP predictions for T0097---no consensus: 7 1.7.1 5 1.75.1 5 1.70.1 5 1.36.1 5 1.35.1 3 3.31.1 3 1.7.5 3 1.7.2 3 1.19.1 2 4.35.3 2 4.108.1 2 3.20.1 2 1.48.2 2 1.4.1 2 1.3.1 2 1.26.1 2 1.25.9 2 1.25.1 2 1.24.2 2 1.105.1 2 1.1.1 About all people agree on it that this is a helical protein. Our previous attempts are all over the place also 1aj8a 1.98.1 1bazA 1.45.1 1bdvA 1.45.1 1c1gA 1.8.1 1a4mA 1.3.1 SAM-T99 got 1.105.1 1sig (though with E value 25.43) The 1sig/T0097-1sig-local alignment gets a good match on the helix RQAILKQGQDGLSGVKETDK, and a nice match on the turn and helix SELARISKLI, but the helix between them has a large insertion relative to 1sig. The global alignment of the 1sig.stride-mixed and 1sig.stride sequences is different, making it difficult to look at the alignment with see-a2m. The 1sig/1sig-T0097-fssp-global alignment scatters the helices in a non-compact arrangement. It might be good as a source of fragments to assemble, but the fragments are much more than single helices. The most pouplar target is 1.7.1, represented by 1quuA, 2spcA, and 1cunA in FSSP. These are coiled-coils. Top-scoring alignments are: 1quuA/T0097-1quuA-global.pw.dist:1quuA 248 -75.17 -10.89 5.6e-05 1a4mA/T0097-1a4mA-local.pw.dist:1a4mA 349 -14.96 -6.00 7.4e-03 1aj8A/T0097-1aj8A-local.pw.dist:1aj8A 371 -16.33 -5.84 8.7e-03 1sig/T0097-1sig-local.pw.dist:1sig 339 -16.70 -5.71 1.0e-02 1a4mA/T0097-1a4mA-vit.pw.dist:1a4mA 349 -9.10 -5.24 1.6e-02 1sig/T0097-1sig-global.pw.dist:1sig 339 -90.78 -4.87 2.3e-02 1aj8A/T0097-1aj8A-vit.pw.dist:1aj8A 371 -8.72 -4.81 2.4e-02 1sig/T0097-1sig-vit.pw.dist:1sig 339 -10.70 -4.62 2.9e-02 2spcA/T0097-2spcA-vit.pw.dist:2spcA 107 -7.98 -4.22 4.4e-02 2spcA/2spcA-T0097-fssp-global.pw.dist:T0097 105 -17.80 -2.78 1.7e-01 : C-terminal domain of ERp29, rat 1cunA/1cunA-T0097-fssp-global.pw.dist:T0097 105 -29.25 -2.44 2.4e-01 : C-terminal domain of ERp29, rat 2spcA/T0097-2spcA-local.pw.dist:2spcA 107 -12.78 -2.29 2.8e-01 1cunA/T0097-1cunA-vit.pw.dist:1cunA 213 -5.96 -1.57 5.2e-01 1quuA/1quuA-T0097-fssp-global.pw.dist:T0097 105 -35.88 -0.61 1.1e+00 : C-terminal domain of ERp29, rat 1cunA/T0097-1cunA-local.pw.dist:1cunA 213 -11.16 -0.45 1.2e+00 1quuA/T0097-1quuA-local.pw.dist:1quuA 250 -12.06 -0.26 1.3e+00 1sig/1sig-T0097-fssp-global.pw.dist:T0097 105 -19.02 0.03 1.5e+00 : C-terminal domain of ERp29, rat 1aj8A/T0097-1aj8A-global.pw.dist:1aj8A 371 -87.07 0.09 1.6e+00 1quuA/T0097-1quuA-vit.pw.dist:1quuA 250 -6.47 0.09 1.6e+00 2spcA/2spcA-T0097-global.pw.dist:T0097 105 -37.42 0.49 1.9e+00 : C-terminal domain of ERp29, rat The 1quuA/T0097-1quuA-global alignment has breaks in the helices in awkward places. The 2spcA/2spcA-T0097-fssp-global alignment looks pretty good, with the NK insertion (predicted to be coil) occuring at the kink in the helix in 2spcA. The alignment 2spcA/T0097-2spcA-vit is much less convincing, since it has a big gap in the helix. 1cunA/1cunA-T0097-fssp-global spreads t97 into discontinuous matches---just helical fragments again. Mon Jun 26 09:52:20 PDT 2000 Remade 2ry predictions Tue Aug 8 11:48:37 PDT 2000 Kevin Karplus make 2track gets weak hits: Sequence Length Simple Reverse E-value FSSP SCOP 1ncx 162 -25.82 -7.33 3.1e+00 1ncx 1.41.1 1elrA 131 -36.89 -7.22 3.1e+00 1a17 1.110.8? 1a4sA 503 -28.84 -6.89 8.5e+00 1bpwA 3.76.1 1bpwA 503 -28.23 -6.72 8.5e+00 1bpwA 3.76.1 1qsdA 106 -34.53 -6.35 8.5e+00 1qsdA 1.7.5 1dgfA 497 -25.20 -6.26 8.5e+00 1cf9A 5.5.1 1aj8A 371 -26.15 -5.91 2.3e+01 1aj8A 1.98.1 2olbA 517 -17.55 -5.66 2.3e+01 1jevA 3.89.1 1pbwA 216 -29.48 -5.63 2.3e+01 1pbwA 1.108.1 1jevA 517 -17.55 -5.63 2.3e+01 1jevA 3.89.1 1jetA 517 -17.55 -5.63 2.3e+01 1jevA 3.89.1 1by1A 209 -27.68 -5.53 2.3e+01 1by1A 1.83.1 1bgf 124 -30.58 -5.45 2.3e+01 1bgf 1.86.1 1apa 266 -18.68 -5.14 2.3e+01 1mrj 4.143.1 Note: e-values appear to be computed incorrectly. The prediction is all helical, so the 3.* and 4.* domains are strange hits. Looking at the top alignments for a few of these: 1quuA/T0097-1quuA-global 1quuA 250 -75.56 -10.97 5.2e-05 some helical pieces, but no matches for turns, and coil predicted in mid helix. 1a4mA/T0097-1a4mA-local 1a4mA 349 -14.96 -6.00 7.4e-03 9 identical residues, but missing beta sheet that separates the helices. An unlikely match. 1ncx/T0097-1ncx-2track-local 1ncx 162 -25.04 -6.94 7.4e-03 6 identical residues, but no match to the distinctive parts of the fold. An unlikely match. 1aj8A/T0097-1aj8A-local 1aj8A 371 -16.33 -5.84 8.7e-03 14 identical residues, with pretty good 2ry structure conservation. This looks like the best I've seen yet, though it doesn't cover that last two helices. 1sig/T0097-1sig-local 1sig 339 -16.70 -5.71 1.0e-02 12 identical residues, only ok 2ry match. Fragment needs more aligned to be compact. 1a4mA/T0097-1a4mA-vit 1a4mA 349 -9.10 -5.24 1.6e-02 tiny fragment---irrelevant. 1ncx/T0097-1ncx-2track-global 1ncx 162 -6.64 -5.54 2.0e-02 distinctive features omitted from target---an unlikely match. 1sig/T0097-1sig-global 1sig 339 -90.78 -4.87 2.3e-02 Only 3 identical residues---and predicted helices broken in unlikely places. 1aj8A/T0097-1aj8A-vit 1aj8A 371 -8.72 -4.81 2.4e-02 looks essentially the same as T0097-1aj8A-local 1sig/T0097-1sig-vit 1sig 339 -10.70 -4.62 2.9e-02 2spcA/T0097-2spcA-vit 2spcA 107 -7.98 -4.22 4.4e-02 2spcA/2spcA-T0097-fssp-global T0097 105 -17.80 -2.78 1.7e-01 13 identical residues, ok 2ry prediction. coiled-coil dimer with turn. Most of the conservation on one part of helix. Matches full length. 1cunA/1cunA-T0097-fssp-global T0097 105 -29.25 -2.44 2.4e-01 2spcA/T0097-2spcA-local 2spcA 107 -12.78 -2.29 2.8e-01 1elrA/T0097-1elrA-2track-local 1elrA 128 -12.01 -2.99 3.6e-01 4 identical residues, just gets two helices with a turn between them. CAFASP fold summary: fold #server #found #first 1.7 9 13 1 1.36 5 5 5 1.35 5 5 4 1.25 5 5 2 1.75 5 5 0 1.70 5 5 0 The structure is MONOMERIC, which makes the 2spcA hit rather unlikely, since that is a dimeric colied-coil. Right now I favor T0097-1aj8A-local. 13 August 2000 Christian I concur about T0097-1aj8A-local, althoug I'm willing to bet that the helices are a bit closer together (4-helical bundle). 14 August 2000 Rachel alignments on some of the weak 2track hits: Nothing good so far 1bpwA/T0097-1bpwA-2track-global.pw.dist:1bpwA Good secondary structure match except for first few residues which are garbage. 1bpwA/T0097-1bpwA-2track-local.pw.dist:1bpwA Very similar to 2track-global 1bpwA/1bpwA-T0097-vit.pw.dist:T0097 Poor secondary structure match. Not compact. 1qsdA/T0097-1qsdA-2track-local.pw.dist:1qsdA Poor secondary structure match. Not compact. 1qsdA/1qsdA-T0097-local.pw.dist:T0097 Didn't align to anything. Really awful. 1dgfA/T0097-1dgfA-vit.pw.dist:1dxgfA Only aligns a single helix. Not compact. 1dgfA/T0097-1dgfA-local.pw.dist:1dgfA Aligns a single helix-turn-helix with two conserved residues. 2olbA/T0097-2olbA-vit.pw.dist:2olbA Seven conserved residues, only part of the target is aligned with a gap in the middle of a helix. Template has sheet where our prediction is helix. Tue Aug 15 02:07:27 PDT 2000 Kevin Karplus Trying undertaker on collections of alignments and fragments. Tue Aug 15 09:52:53 PDT 2000 Kevin Karplus The first two tries with undertaker yielded quite different structures, neither particularly well packed. Both structures were all helical, as our secondary structure prediction expects, but our prediction had longer helices. It may be necessary to add a new operator to undertaker---one that inserts helical or extended conformations where those are predicted. Sat Aug 26 15:44:33 PDT 2000 Kevin Karplus Remade 2track predictions % Sequence ID Length Simple Reverse E-value X count 1ncx 162 -25.94 -7.41 3.1e+00 1elrA 131 -35.62 -7.04 3.1e+00 1a4sA 503 -28.15 -6.98 8.5e+00 1bpwA 503 -27.53 -6.86 8.5e+00 1dgfA 497 -25.04 -6.25 8.5e+00 1by1A 209 -28.22 -5.99 2.3e+01 1qsdA 106 -31.81 -5.88 2.3e+01 1pbwA 216 -29.48 -5.57 2.3e+01 2olbA 517 -16.67 -5.57 2.3e+01 1jetA 517 -16.67 -5.54 2.3e+01 1jevA 517 -16.67 -5.54 2.3e+01 1apa 266 -18.90 -5.35 2.3e+01 1bgf 124 -30.23 -5.24 2.3e+01 1a1z 91 -19.89 -5.21 2.3e+01 1.75.1 Only the last of these is new to this search.