Flowchart



Examples

BIGS protein YecD (1J2R) : an E. coli protein belonging to the isochorismatase family
The Escherichia coli gene yecD encodes a 20 kD protein of unknown function annotated in Swiss-Prot as belonging to the isochorismatase family (Swiss-Prot: P37437). The crystals belong to the orthorhombique space group P21212 with 4 molecules in the asymmetric unit. The YecD protein shares 19.5% and 24.6% identity with the 1IM5 and the 1NBA structures. Out of the 60 models generated and screened through the CaspR procedure, 30 correspond to truncated models (positions 1 to 18, 35 to 39 and 68 to 103). In the sorted list of the best ranking solutions in AMoRe, the 1NBA template is 13th and 1IM5 is 16th. The ten best solutions all correspond to truncated models and all converged through the pre-refinement process while 1NBA and 1IM5 did not (best solution: Rfree: 50.6, Rwork: 42.0; 1NBA: Rfree: 58.0 Rwork: 56.5; 1IM5: Rfree: 59.6 Rwork: 57.1). This case is actually the first structure that has been solved entirely using the method now automated in CaspR.
To look at the general result page click here
To look at the T-COFFEE alignment click here
To look at the AMoRe result page click here
To look at the CNS result page click here
1AJX: the HIV1 protease
The HIV1 protease structure belongs to the acid proteases fold. It is a homodimer of a 99 residue long protein and can adopt two conformations upon ligand binding (RMSD between 1.0 and 1.2 Å over 99 residues based on C. superimposition of the two structural conformations). Each of these structures has been solved independently and correspond to 1AJX (1) and 1HHP (2) PDB accession numbers. We used the available 1HHP structure to solve the 1AJX structure with CaspR. In this example, the crystals belong to the orthorhombic space group P21212 and there are 2 molecules in the asymmetric unit. While the 1HHP structure clearly identifies the first position in the AMoRe step of the CaspR process, it fails to identify the second molecule using the standard procedure. On the other hand, using the CaspR automated protocol, 10 out of the 30 models produce a two molecule solution in AMoRe and converge during the refinement process (best model Rfree: 38.4, Rwork: 31.4; 1HHP Rfree: 54.8, Rwork: 50.6).
To look at the general result page click here
To look at the T-COFFEE alignment click here
To look at the AMoRe result page click here
To look at the CNS result page click here
1K6D : the E. coli . subunit of the acetate CoA-transferase
The 1K6D structure (3) belongs to the CoA transferase fold. It crystallizes in the hexagonal space group P62 with two molecules in the asymmetric unit. This 220 residues long protein has a closely related structural homologue exhibiting 35% sequence identity over its entire length (1M3E PDB accession number) as well as a more distant homologue with less than 25% sequence identity (1POI PDB accession number). Both potential templates were tested using the CaspR web-server. In the case of 1M3E, 30 models were produced by MODELLER and residues that were poorly aligned to the target were automatically excised based on the T-COFFEE core index (residues 1 to 7, 124 to 134 and 216 to 220). As a result, the 14 best AMoRe solution were obtained using the truncated models. 28 out of the 30 truncated and 21 out of the 30 non truncated ones produced an AMoRe solution for both molecules in the a.u.. All of them present a higher correlation coefficient and a lower R-factor in AMoRe than the unperturbed 1M3E structure. The 10 best solutions all correspond to the truncated models and converge during the pre-refinement procedure using CNS (best solution Rfree: 43.7, Rwork: 36.3; 1M3E: Rfree: 56.9 Rwork: 53.2)
In the second case (1POI), 60 models were generated, 30 of them corresponding to the truncated forms based on the T-COFFEE core index (excision of residues 58 to 60 and 218 to 220). In this case, the best AMoRe solution is given by the 1POI structure itself, while none of the solutions converged during the pre-refinement process. This is a case where CaspR was not able to improve the solution with a homology model. Neither the models nor the related structure could produce a result in this standard procedure. A way to retrieve a solution may be to change the core index threshold value to be used in the design of truncated models.
To look at the general result page (1M3E) click here To look at the general result page (1POI) click here
To look at the T-COFFEE alignment (1M3E) click here To look at the T-COFFEE alignment (1POI) click here
To look at the AMoRe result (1M3E) page click here To look at the AMoRe result (1POI) page click here
To look at the CNS result (1M3E) page click here To look at the CNS result (1POI) page click here
1MP0: The human glutathione-dependent formaldehyde dehydrogenase
The 1MP0 structure (4) is a two domains structure belonging to the GroES-like fold for the first domain and to the NAD(P)-binding Rossmann-fold for the second domain. The crystals belong to the tetragonal space group P43212 and there are 2 molecules of this 373 residues long protein in the asymmetric unit. A closely related structure (1N8K) sharing 56% sequence identity over 373 residues as well as more distantly related one (1JQB and 1JVB) sharing less than 26% sequence identity with 1MP0 were tested through the CaspR web-server.
In the first case (1N8K), 60 models were produced by MODELLER with 30 truncated models based on the T-COFFEE core index (excision of residues 1 to 5 and 245 to 251). The best solution in AMoRe correspond to the 1N8K structure itself. However, all truncated and non truncated models also produced a solution which converged through the pre-refinement procedure using CNS. Eventually, the 10 best solutions in CNS all correspond to homology models and score higher that 1N8K itself, most likely because the models carry the side chains that are proper to 1MP0 (best solution Rfree: 41.4, Rwork: 35.4; 1N8K: Rfree: 48.8 Rwork: 49.1. This shows that even in .easy. MR cases, CaspR is still helpful by improving the starting model prior to manual refinement.
In the second case, 60 models were produced by MODELLER based on the two 1JVB and 1JQB structures. The 30 truncated models correspond to the N-terminal removal of the 4 first residues of the 1MP0 sequence. Both the probe structures and the models failed to produce a convincing solution in AMoRe and did not converge in the CNS refinement procedure. As for 1K6D, the lowering of the T-COFFEE core index threshold may be a way to generate new truncated models for solution screening in AMoRE. Molecular replacement can also be run using the two sub-domains independently for the model generation and the two-body screening for solutions in the AMoRe step.
To look at the general result page (1N8K) click here To look at the general result page (1JQB/1JVB) click here
To look at the T-COFFEE alignment (1N8K) click here To look at the T-COFFEE alignment (1JQB/1JVB) click here
To look at the AMoRe result (1N8K) page click here To look at the AMoRe result (1JQB/1JVB) page click here
To look at the CNS result (1N8K) page click here To look at the CNS result (1JQB/1JVB) page click here
BIGS protein YhbO (1OI4): The E. coli potential cysteine peptidase protein
YhbO corresponds to a 193 residues long protein annotated as a potential cysteine peptidase (Swiss-Prot: P45470). It belongs to the Flavodoxin-like fold and shares 41% identity with its structural homologue 1G2I. The crystals belongs to the orthorhombic space group P212121 with 2 molecules in the asymmetric unit. CaspR produced 60 models corresponding to 30 truncated forms (excision of residues 1 to 24; 56 to 63 and 168 to 171) which went through the molecular replacement procedure using AMoRe. The best scoring AMoRe solution is given by the 1G2I structure itself. However all 60 models also produce a correct solution. After refinement procedure with CNS, the 1G2I solution eventually only ranks third below two truncated models with a better convergence, which again can be explained by the replacement of the 1G2I sequence by the proper YhbO one (best solution Rfree: 44.9, Rwork: 37.5; 1G2I: Rfree: 45.7 Rwork: 38.4). This again is a case were CaspR is helpful to decrease the time spent in refinement procedure.
To look at the general result page click here
To look at the T-COFFEE alignment click here
To look at the AMoRe result page click here
To look at the CNS result page click here
BIGS protein YahK (1UUF): The E. coli Zinc-type alcohol dehydrogenase-like protein
YahK is a 349 residue long protein annotated as a Zinc-type alcohol dehydrogenase-like protein (Swiss-Prot: P75691). Like 1MP0, it is a two domain structure belonging to the GroES-like fold for the first domain and to the NAD(P)-binding Rossmann-fold for the second domain. It shares 33% identity with its closest homologue 1LLU and 27.6% and 26.5% identity with the two related structures 1H2B and 1JVB. The crystals belongs to the monoclinic space group C2 and there is one molecule in the asymmetric unit. Out of the 60 models produced using the 3 template structures through the CaspR process, 30 correspond to truncated forms based on the T-COFFEE core index (10 excised segments corresponding to positions 1 to 48, 72 to 80, 110 to 115, 129 to 146, 156 to 168, 191 to 193, 248 to 260, 290 to 304, 327 to 336 and 364 to 369). Both 1LLU and 1JVB produces higher scoring solution in the AMoRe search, however, the best solution in the pre-refinement procedure is produced by one of the 30 non truncated models (best solution Rfree: 49.9, Rwork: 40.7; 1LLU: Rfree: 51.1 Rwork: 42.0). Due to their ranking, none of the truncated models went through the refinement process.
To look at the general result page click here
To look at the T-COFFEE alignment click here
To look at the AMoRe result page click here
To look at the CNS result page click here
BIGS protein YggV (also available as 1K7K) : The E. coli HAM1 NTPase protein
YggV is a 197 residue long protein belonging to the HAM1 family of pyrophosphatase (Swiss-Prot: P52061). According to the SCOP database it belongs to the Anticodon-binding domain-like and shares 33.5% identity with the related 2MJP structure. The crystals belong to the tetragonal space group P43212 and there is one molecule in the asymmetric unit. 30 full length models having been produced through the automated procedure and generated 30 truncated models based on T-COFFEE core index (excision of segments 22 to 39, 89 to 111, 119 to 133 and 191 to 197). In AMoRe, the 2MJP template structure is ranked in position 27th in the list, while the best ranking solutions corresponds to the full length models and after the pre-refinement process, 2 full length models comes first and are immediately followed by the 2MJP template structure (best solution Rfree: 54.3, Rwork: 44.7; 2MJP: Rfree: 55.3 Rwork: 45.2). The amplitude of the R-factors and difference between that of the free and the working set prompt for a visual inspection of the corresponding electron density maps, based on for highest scoring model and on 2MJP itself, in order to validate the CaspR solution(s). It turned out that sole the solution based on the homology model provided a usable map for further refinement (most likely due to the presence of the correct side chains in the model).
To look at the general result page click here
To look at the T-COFFEE alignment click here
To look at the AMoRe result page click here
To look at the CNS result page click here
REFERENCES
1.Backbro, K., Lowgren, S., Osterlund, K., Atepo, J., Unge, T., Hulten, J., Bonham, N. M., Schaal, W., Karlen, A. & Hallberg, A. (1997) Unexpected binding mode of a cyclic sulfamide HIV-1 protease inhibitor. J. Med. Chem., 40, 898-902.
2.Spinelli, S., Liu, Q. Z., Alzari, P. M., Hirel, P. H. & Poljak, R. J. (1991) The three-dimensional structure of the aspartyl protease from the HIV-1 isolate BRU. Biochimie, 73, 1391-1396.
3.Korolev, S., Koroleva, O., Petterson, K., Gu, M., Collart, F., Dementieva, I. & Joachimiak, A. (2002) Autotracing of E. Coli Acetate Coa Transferase A-Subunit Structure Using 3.4 A MAD and 1.9 A Native Data. Acta Cryst., D58, 2116-2121.
4.Sanghani, P. C., Robinson, H., Bennett-Lovsey, R., Hurley, T. D. & Bosron, W. F. (2003) Structure-Function Relationships in Human Class III Alcohol Dehydrogenase (Formaldehyde Dehydrogenase). Chem. Biol. Interact., 143, 195-200.